Transcript Construction & Disto..

Construction &
Distortion of Memories

[This page is intentionally left blank.]
Memory and Gist

Gist ~ theme ~ schema ~ expectation

Fuzzily defined, not identical, but clearly related

Memory is designed to operate on gist, it can function
without gist (as Ebbinghaus showed), but gist is what it
does best.

We think of memory as a series of processes (encoding,
organizing, storing, retrieving, etc.) and gist influences
every stage.
Memory and Gist

Bransford, J.D., & Johnson, M.K. (1973) reasoned that, if
memories are the sum of experience and knowledge, then
it is possible that a person first comprehends an input and
then elaborates on it. If this is true, then people should
remember text equally well if relevant knowledge is
unavailable.
Memory and Gist

They told S’s to listen carefully and try to understand a
passage, and that they would be asked for immediate
recall. They manipulated the availability of the relevant
information by showing (or not) the cartoon for 30 seconds
prior to hearing the passage. Comprehension ratings were
gathered as a manipulation check to determine whether
this effected the ability to extract gist. The manipulation
worked: those shown the cartoon understood more and
remembered more.
Memory and Gist

Isn’t it possible that people remembered more simply
because they had more opportunities to encode the
material (with a picture vs. no picture)?

A series of replications were run to eliminate other
possibilities. A second no context group got to hear the
passage twice in a row. An after group was shown the
cartoon after having heard the passage. A partial context
group was shown a related picture.
Memory and Gist
Memory and Gist
Mean Comprehension Ratings and Mean Number of Ideas Recalled
No
No
Contxt Partial Contxt
Contxt1 Contxt2 After Contxt Before
Comprehension 2.30
Recall
3.60
3.60
3.80
3.30
3.60
3.70
4.00
6.10
8.00
The mean comprehension ratings and mean number of ideas recalled
correlate highly. (r=0.94)
Memory and Gist

In a subsequent replication an additional factor was added,
free recall vs. cued recall.
Mean Number of Ideas Recalled (Balloon Passage)
No Cue
Key Word Cues

No Contxt
3.92
4.00
Contxt After
4.33
3.75
Cntxt Before
7.33
8.50
The cues don’t make much difference for the noncomprehending groups, because most of the information
just isn’t there to be retrieved.
Memory and Gist

The moral of the story is that gist drives encoding.
Memory and Gist
Memory and Gist

Brewer & Treyens (1981) created a false “graduate student
office”, containing 61 objects designed to vary along two
dimensions:


more or less noticeable in the room (saliency)
more or less likely to exist in a GSO (schema-expectancy).

Had two groups of subjects rate the objects in the office
(one group for each scale) and 70 additional objects that
were not present in the office.

Saliency and schema-expectancy correlated -.41 for
present objects and -.69 for lures.
Memory and Gist

S’s were asked to wait in the room for 35 seconds, then
taken to a plain room and told about the experiment. S’s
were split into three groups that were given different
memory tasks: written recall, drawing recall on provided
outlines, and recognition. The frequencies with which
objects were listed in the recall tasks was tabulated.
Correlation between the two recall measures was .94.
Memory and Gist
Partl Cors b/t Recall Freq & Schema Expectancy & Saliency
SE
Sal
Written Recall Frequency
.55
.64
Drawing Recall Frequency
.68
.66
Partl Cors b/t Mean Recognition & SE & Sal (recognzd items)
SE
Sal
Present Object Recognition
.58
.69
Absent Object Recognition
.52
-.36
Memory and Gist

Both things that we expect to see and those that, for
whatever reason, we are surprised to see (that stand out)
tend to get noticed and remembered. This was supported
by qualitative analyses. The study demonstrates
expectancy congruent and expectancy incongruent
encoding.

But only things we expect lead to intrusions (expectancy
congruent distortion on retrieval).
Memory and Gist

Some S’s performed both recall and recognition, so Brewer
& Treyens created a retrieval ratio by dividing the number
who recalled an object by the number that gave it the top
recognition rating. This ratio doesn’t correlate significantly
with saliency, but partial correlation between retrieval ratio
and SE was .56.

This shows that the schema (the gist) facilitates retrieval.

Similar results found by Hastie & Kumar (1979):
inconsistent highest, consistent high, neutral lowest, p.166
in Kunda.
Memory and Gist

Owens, J., Bower, G. H., & Black, J. B. (1979) had S’s read
passage first day and estimated time.

P group estimated significantly less reading time, despite
having read more.

Recalled as much as possible the next day.


P group remembers more episodes, more often in the right order,
more propositions, but also more intrusions (false alarms).
Then cued with titles of episodes for additional recall.

C group remembers more and more accurately.
Memory and Gist
Recall Measures for Prblm and Cntrl Conditions Before & After Cuing
Scripts
Recalled*
Props
New Props Total Props
Recalled* Intruded** Recalled
B/f Cue Problem
Control
3.67
2.50
29.24
20.24
15.20
3.76
44.44
24.00
A/f Cue Problem
Control
4.56
4.75
38.44
54.63
17.56
8.88
56.00
63.51
Condition
* Out of 5
**Out of 190
Memory and Gist

Then test for recognition using true propositions, and lures
of various types, using 7 point Likert scale ratings.


C group more accurate on trues, and P group less accurate on motive
relevant lures.
Lesson: a theme (or gist) organizes memories and
facilitates retrieval. Episodes that lack gist are not
necessarily gone, but are disorganized and very difficult to
retrieve.
Memory and Gist

Anderson, R. C., & Pichert, J. W. (1978) wanted to
investigate whether storage processes are distinct from
retrieval processes.
Memory and Gist

Created a story about two boys staying home and playing
hooky. The story contained information about the house
that would be pertinent to a prospective homebuyer or a
burglar. The ideas relevant to each were assessed in
advance with ratings.

S’s were told to read the story from one of the two
perspectives. They performed a distracter task for 12
minutes, and were asked to recall as much of the story as
they could. S’s then performed a second distracter task,
and were asked to recall the story again, but this time from
the other perspective.
Memory and Gist
Proportion of Idea Units Recalled a Function of Perspective
1st/2nd Perspctve
Homebuyer/burglar
Burglar/homebuyer

Burglar
---------------------1st Recl 2nd Recl
.51
.68
.61
.36
Homebuyer
--------------------1st Recl 2nd Recl
.59
.40
.48
.50
What’s important here is that on the second recall S’s remembered more
ideas relevant to the new perspective than they had initially. Interviews
after the experiment supported this result.
Memory and Gist

The big idea is that a theme can organize memories and
facilitate recall even if introduced after the memory has
been encoded (i.e. independently of encoding and storage).
Source Amnesia, Source Monitoring
and The Hindsight Bias

First, an memory test
Source Amnesia, Source Monitoring
and The Hindsight Bias
Well, how did you all do?
 Any false positives?
 Kunda (ch 4) discusses that we often
mistake the memory for something we
imagined for something that actually
happened.
 As imagined events become more rich, we
become more likely to mistake them for
real events.

Source Amnesia, Source Monitoring
and The Hindsight Bias
Roediger & McDermitt (2000)
 The words on the list activated their
central theme.
 The central theme is strongly remember,
although it was not presented.
 Was the central theme explicitly or
implicitly activated?
 There are data to suggest it was explicitly
activated, what do you think?

Source Amnesia, Source Monitoring
and The Hindsight Bias
Alzheimer’s patients are much worse at
recalling words in list, but their false
positive rate is the same as healthy
adults.
 Does anyone know other examples of
people with memory loss who maintain
their ability to encode and retrieve central
themes but lose the periphery?

Source Amnesia, Source Monitoring
and The Hindsight Bias
Kunda also discusses stereotype driven
inferences as a source of source amnesia.
 People here that “the lawyer is standing in
front of her house.”
 Later, people report that they remembered
hearing about a mansion.

Source Amnesia, Source Monitoring
and The Hindsight Bias



People may confuse where there heard
something when the two potential sources are
seen as similar to each other, i.e. confusing which
of the two female justices.
Dimensions that cause us to categorize change
depending on our theories of the world.
One’s dress only lead to source monitoring
problems when that dimension was important,
i.e. when deciding if someone should be a media
rep.
Source Amnesia, Source Monitoring
and The Hindsight Bias
Hard to reconstruct passed predictions
after the event occurs.
 Monday Morning Quarterbacks.
 When we retrieve info we reconstruct
using current knowledge, cannot escape
this.
 People estimate that others who do not
know the result will predict the result
because it is obvious.

Source Amnesia, Source Monitoring
and The Hindsight Bias
Mather, Shafir, and Johnson (2000)
 Relates to Source Monitoring and
Hindsight Bias (a little).
 When remembering past choices, people
engage in choice supportive memory
distortion.

Source Amnesia, Source Monitoring
and The Hindsight Bias
Choose what roommate you would prefer
between two people with positive and
negative features.
 After choice, subjects see list of features
and say whether the feature went with
roommate A, B, or it’s new.
 Some of the positive features of B are
misattributed to A (fewer negative
features are misattributed).

Source Amnesia, Source Monitoring
and The Hindsight Bias
When did misattribution occur?
 At first encoding causing one choice to
seem better causing the choice?
 When the choice was made to ensure later
remembering congruent with a good self
image?
 At the time of source monitoring?
 Positive new features misattributed to A,
at least some happened at final stage.

Source Amnesia, Source Monitoring
and The Hindsight Bias
Why are we designed to err?
 Flexibility, powerful inductive abilities and
maintaining a good self-image has it’s
price: maintaining stereotypes, etc.
 Are you happy with your memory system?

Eyewitness Testimony
PROS
CONS
Eyewitness Testimony
PROS
Ecologically
valid research
Lowers false conviction rates
CONS
Eyewitness Testimony
PROS
CONS
Age
Person
perception
Emotional interference
Confidence in confidence
Source amnesia
Misinformation effect
Eyewitness Testimony
Misinformation Effect


Overwriting
Interference
Eyewitness Testimony
PROS
Ecologically
valid research
Lowers false conviction rates
CONS
Age
Person
perception
Emotional interference
Confidence in confidence
Source amnesia
Misinformation effect
Eyewitness Testimony
If we suck so bad at it, …
should we ditch it?