Interconnections between

1
Chapter 5

• Interconnections between
• Acquisition Retrieval

2
The Important of Retrieval
As highlighted in the previous chapter, storing
information in memory is useless if that
information cannot be efficiently retrieved at a
later time. We have highlighted the importance
of depth of processing with respect to making
things retrievable, loosely talking in terms
of the importance of making connections with
other information. In this chapter, we will
discuss this issue in more detail,
focusing specifically on the link between how
information is acquired and how it is
subsequently retrieved.
3
A Running Theme
There will be a familiar theme to many of the
results we are about to discuss the theme is
this what you do at encoding strongly effects
how well you are able to subsequently
retrieve. Well discuss several issues relevant
to this theme including 1) matches between the
expectations of the participants and the
eventual test 2) matches between the context of
the study and the context of the test
(state dependent learning) 3) the general
notion of encoding specificity
4
Recall versus Recognition
So far, we have focussed mostly on memory results
gathered using a free recall task one where
participants are first shown a list of items,
then are asked to recall as many of them as
possible. We will now begin to discuss results
from a different kind of test, one called
old/new recognition. Here participants are
again first shown a study list of items.
However, their memory is tested differently.
Specifically, items are presented one at a time,
typically with half of the items being from the
study list, and half not. The participants is
then asked to give an old or new response to
each item where old means they think the item
was on the study list, and new means they think
it was not on the study list.
5
An Example - Study List
TRUCK PINK TREE SPEAKER PHONE SHARK HAMMER UMBRELL
A DISK RECORD CLOCK FACE PAPER STAPLER PLAQUE CAND
LE CACTUS DOG COUCH WIRE ENVELOPE
6
An Example - Test Phase
Free Recall write down as many of the words as
you can recall Old/New Recognition give an
old or new response to each of the following
items HORSE RECORD STAPLER HAMMER NOSE PAPER
HEALTH TOOTH CLOCK TISSUE DISK BASKET UMBRELLA
DRILL PAD FACE SLEEP PLAQUE CHAIR MILK
7
Score Yourself
Free Recall How many of the green words listed
below did you recall? Old/New Recognition How
accurate where you? Hits versus false
alarms. HORSE RECORD STAPLER HAMMER NOSE PAPE
R HEALTH TOOTH CLOCK TISSUE DISK BASKET UMBREL
LA DRILL PAD FACE SLEEP PLAQUE CHAIR MILK
8
Differences in the Tests
As you likely noted, there two ways of testing
are not the same for free recall, you have to
come up with the items yourself at test, and
sometimes that seems difficult in and of itself
(like short answer). In old/new recognition, the
items are provided to you so you do not have to
come up with them rather, you just have to
recognize whether or not a given item was on the
study list this generally makes it an easier
test (like multiple choice tests). Would it
bother you if I told you that the exam was
multiple choice, then it turned out to be short
answer? What about the reverse? If you know the
material, would it matter what kind of test I
gave?
9
Expectations Confirmed or Denied
An experiment by Tversky (1973) examined just
this issue. In that experiment, participants
were told, prior to viewing the study list, that
they were going to get either a free recall test,
or an old/new recognition test. Then, half of
the participants did receive the test they were
told to expect (that is, some were told to
expect free recall and did, others were told to
expect old/new recognition and did). The other
half of the participants did not receive the test
they were told to expect (that is, some were
expecting free recall but got old/new recognition,
others were expecting old/new recognition and
got free recall) predictions from the audience?
10
Results
Participants did best on the recall test when
they were expecting recall and they did best on
the recognition test when they were
expecting recognition. Also note that they
generally did better with the recognition test
it is easier.
11
The Story
Why do participants do best when they get the
kind of test they expect? The story would go
like this. When they are studying, they are
aware of how they will be tested, and they encode
the information in a way that will make it easy
to retrieve in the test context. Imagine I had
to remember the light spectrum for a test (Roy G
Biv), how would my study be different for a
multiple choice versus a fill in the blank
test? The moral is that we tailor our encoding
to the expected retrieval context and when we
are right, we retrieve better. But what
underlies this phenomenon? To get at this, lets
consider more data ...
12
State Dependent Learning
There have been a large number of studies that
all address a similar question using a similar
experimental design. The question is - does a
match between the study and test context lead to
better memory? The general design then involves
having participants study some items in context 1
or 2, then getting them to retrieve them in
context 1 or 2 if a match matters then we
expect the best performance when the study and
test contexts are the same (1-1 or 2-2) and
worse performance when they differ (1-2 or 2-1).
13
A Concrete Example
An examination of state-dependent learning with
SCUBA divers
Divers saw the study list either on land
or underwater, then had their memory
tested either on land or underwater. There are
funky water- proof slates you can use to do
this.
14
Results
Participants who learned on land did best when
tested on land, and those who learned underwater
did best when tested underwater. This is likely
not due to any differences in the way items
were encoded but, instead, seems to suggest that
the context in which the items were presented is
encoded with the items and when the
test context matches this, it helps retrieval
15
Internal States
This kind of effect does not appear limited to
the external context in which the material
occurred, internal context also appears
relevant. For example, consider a drug
experiment conducted by Eich et al.
(1975). Apparently, things learned while
stoned are best remembered while stoned.
16
Even Emotions
There is some evidence, though it is not as firm,
that even matches in emotional states might
influence memory. For example ...
Bower (1981) reported the data shown opposite
where things learned while sad were
best remembered when sad, and things learned
while happy were best remembered while happy.
17
Tying this all back to Depth
Last chapter I claimed that deeper processing
leads to better memory. Now we seem to be saying
that the best memory occurs when the test
context (internal or external) matches the study
context. How do these two points come together?
If the test where a shallow test, would shallow
processing be better? That is, does deeper
processing only lead to better memory
because most memory tests are deep tests, or is
there something to depth that goes beyond just a
study-test match? A study by Fisher Craik
attempted to examine this issue directly ...
18
Depth Matching
Fisher Craik (1977) presented participants with
pairs of words and told them to remember the
second word in each pair. Some pairs were
similar in meaning (dog-cat) encouraging
participants to encode the meaning of the studied
word, others were similar in sound (hat-cat)
encouraging participants to encode the sound of
the word. At test, participants were given a
context word and asked things like was their
a word in the list associated with dog? -
meaning was their a word in the list that
rhymed with hat? - sound
19
Results
As indicated below, there was indeed an overall
benefit of encoding in terms of meaning, but this
benefit only really applied to the memory test
that also encouraged meaning processing. In
the sound test, sound encoding actually lead
to superior memory performance
20
Encoding Specificity
All of these results demonstrate a basic finding
that memory is best when the test context matches
the study context, but why is this the case? The
general notion used to explain this is termed
encoding specificity and, according to it, we
encode the context an item occurs it along with
the item in fact, the memory we store is a
combination of the stimulus and its context
thus, in order to retrieve that memory it is
often necessary to reinstate the context if a
different context is used, the item simply will
not be recalled. present OTHER, test with
THE, HE or HER example
21
Two Final Examples
Present GRASP - BABY at study then ask
participants to generate items to words including
INFANT (they generate BABY) then ask if any of
the words they generated were in the study list
they tend to say no to BABY. Present images to
participants that are all white figures on black
backgrounds, then present the image to the right
later, at test, present that picture again by
first focus them on the black faces was that
image presented at study? Usually they will
say no and be very confident about it.
22
Complicating it all
Up until this point, we have discussed memory is
if it involved a single process of
remembering. That is, we have been assuming that
when participants try to recall something, they
simply search memory to see if they can identify
items that were presented previously. It turns
out, though, that human memory processes are much
more complex and interesting than that and the
rest of this chapter begins to reveal this
complexity.
23
Familiarity vs. Source Recollection
An example to suggest there are at least two
different ways that memory can influence
performance Professor in the Bus
example Items we have recently or repeatedly
experienced tend to feel familiar, where
familiarity is some vague feeling of recent
experience that is sometimes devoid of more
specific information sometimes high levels of
familiarity are enough to help us (e.g.,
recognition memory) Alternately, we sometimes
remember the exact source of the memory influence
we remember where we learned something this
is called source recollection.
24
Recall Revisited
If we consider the recall test in light of this
distinction we would likely conclude that
familiarity likely has little to do with
it. Rather, participants are trying to remember
words from the study list and, to do so, they
likely search memory for words that can recollect
as being on the study list thus it is likely
based primarily on source recollection processes
alone. This claim is supported by the fact that
participants will very seldom provide words that
were not on the list even words that are
very familiar to them.
25
Recognition Revisited
In contrast, the recognition test involves
presenting items to the participants that may or
may not have been on the list. Clearly
participants would likely say old to any item
that they recollected as being on the study
list. However, they might also say old to
items that the do not recollect if those items
are very familiar that is, they might think
that item seems very familiar, it must have been
on the list. This story is supported by the
fact that participants do false alarm a lot to
high frequency (familiar) new items.
26
Familiarity Attribution
At this point we need to be clear about how
familiarity is assumed to effect performance
specifically ... First, it is assumed that ANY
experience with a stimulus allows that stimulus
to be more fluently processed next time it is
experienced (raising a resting threshold?). This
fluency is subjectively experienced as what is
sometimes described as an undifferentiated
familiarity (Ive seen that before). Typically,
we then try to attribute the familiarity to
something and, as youll see, WHAT the
familiarity is attributed is often open
to manipulation.
27
Déjà Vu Example
Imagine the following story 1) Steve takes a
quick glance down the hall and very briefly
sees Big Bird walking his way. 2) However,
before Steve is aware of seeing Big Bird, he
is distracted by a crash behind him. 3) Now
he looks down the hall again and is able to
fluently perceive Big Bird (because of the
head start) leading Steve to think wow,
this feels really familiar man! 4) Steve knows
he has never experienced this stimulus before
not in this life at least!! Creepy man!
28
More on Attribution
The déjà vu example is assumed to represent the
very uncommon situation in which a stimulus is
perceived with a high level of familiarity but
the person has nothing to attribute the
familiarity to. This situation is rare because,
it seems, if we do have any reasonable thing to
attribute the familiarity to, we will do so,
often in error. We have seen one example of this
recognition memory the idea there is that if
a stimulus is familiar (e.g., a very common word)
participants will often false alarm to it this
is thought to be due to them mis-attributing the
familiarity of the item to it being presented on
the study list, not to it being a common word.
More examples ...
29
Becoming Famous Overnight!
Experiment by Jacoby et. al (1989) In Phase 1,
participants are asked to read aloud names
that have been made up by the experiment (these
are not famous names and participants are told
so). --- some delay ensues ----- In Phase 2,
participants are shown a bunch of names, some of
which are famous, some of which are not. Half
of the non-famous names are the same as those
presented in Phase 1.
30
Fame Results
If Phase 2 is run right after Phase 1,
participants will tend to call the Phase 1 names
non-famous because, even though they are
familiar, they know that the familiarity was due
to them being presented in Phase 1 not because
they are famous. However, if Phase 2 occurs a
day after Phase 1, now participants mis-attribute
the familiarity to fame, and they now tend to
call the Phase 1 names famous (hence the
title). Moral source memory seems to decline
quickly and, once it is gone, participants will
tend to attribute familiarity to whatever is
most obvious (in this case fame, since they were
making fame judgments).
31
Worse Still Persuasion
This type of mis-attribution of familiarity can
have scary effects, as it likely occurs (and is
used) frequently in day-to-day life Experiment
by Begg et. al (1992) Participants were read
trivia sentences, and were told that half were
true and half were false. Then were then asked
to judge the credibility of more
trivia sentences, some of which were new, and
some of which were the ones they were
read. They tended to rate the repeated
(familiar) sentences as more credible, even
though they new that half were false.
32
A Step Further
In another experiment, participants read
sentences in which named characters made claims
they were told that the claims made by the male
characters were true, but those made by the
females were false. Frank says that house mice
can run 4 miles per hour Gail says that
crocodiles sleep with there eyes open They were
then given just the topics of the sentences
later, and asked to judge their
credibility. When they remembered who was
associated with each claim, the called the
female ones noncredible (as they should) but
when they did not remember, they judged any old
sentence as credible.
33
STOP HERE FOR NOW!

• The Midterm will cover the information up to this
  point, which corresponds with page 193 in the
  text (up to the Theoretical Treatments of
  Implicit Memory section).

34
Familiarity Implicit Memory
Thus, familiarity is often viewed as providing an
implicit memory influence that is, the
familiarity suggests that the item has
been experienced before despite the fact that the
person feeling the familiarity may have no
explicit memory of that experience. Thus,
experiences may influence our behavior explicitly
when we recollect them or implicitly via
familiarity. Some believe these two types of
memory influence arise completely independently
(I dont, but some do) suggesting that an item
can be familiar without being recollected, both
familiar and recollected, or recollected without
being familiar.
35
Data Supporting this Distinction
Consider the following experiment by Jacoby
(1983) In a study phase participants
experience words in 1 of 3 ways 1) XXXXX -
DARK no context condition 2) LIGHT -
DARK context condition 3) LIGHT -
???? generate condition They are then given
either a recognition memory test, or a
tachistoscopic identification
task. Predictions?
36
Jacoby (1983) Results
Recognition memory shows the typical depth
of processing result, but T-scope
identification shows the reverse pattern
why?
37
T-scope Interpretation
T-scope identification is a perceptually
difficult task. We know that past experience
with a stimulus causes priming on this task (that
is, identification is better for repeated stimuli
- fluent perception). But what sort of
repetition should lead to the best priming?
Probably a repetition of the perceptual form of
the stimulus, given that it is a decision about
perceptual form (and not memory) that is required
in the T-scope task chances are, participants
process the form of the word the most in the no
context condition, and least in the
generate Note that priming is a memory
influence, but memory is being tapped implicitly
(participants are not being asked to remember).
Thus, implicit memory behaves differently from
explicit memory
38
Implicit Memory Awareness
Implicit memory is often viewed as a form of
memory that does not require awareness of the
memory. Reconsider the no context condition of
the Jacoby (1983) experiment, participants
recognition was at 55 chance is 50 clearly
they did not seem to remember much. Nonetheless,
there was substantial priming as evidenced by
their relatively high tachistoscopic
identification performance. Thus, experiences
that we cannot consciously remember, may
still effect our behaviour implicitly.
39
Read the Text!
The text goes into some detail about current
assumptions relevant to implicit memory most of
these things we have discussed throughout the
class here and there. You should read through
those sections, and you are responsible for them,
but I am not going to spend much time on them in
class (a trend a may have to repeat on
occasion). Instead, I will discuss one issue of
interest, and mention some of the relevant
concepts while discussing that issue the issue
is ... AMNESIA!
40
Amnesia Implicit Memory
First off, amnesia comes in two
varieties Retrograde amnesia is the loss of
memory for events that occurred prior to the
onset of amnesia this is often associated with
a good whack on the head though, despite what
you see in movies, is not typically reversed by
another whack. Anteriograde amnesia is the loss
of an ability to lay down new memories. This
is usually caused by a deeper brain trauma, and
an individual who suffers from it would seem
perfectly fine until you asked them about
something that happened yesterday.
41
Causes of Anteriograde Amnesia
A famous case of anteriograde amnesia is the case
of H.M., a young man (at the time) who suffered
from an extreme case of epilepsy and had his
hypocampus and amygdala structures
bilaterally removed. Good news, epilepsy is
fixed bad news, no new memories Anteriograde
amnesia can also be caused by a vitamin
deficiency common to alcoholics (known as
Korsakoffs syndrome) again this syndrome tends
to cause damage to the hypocampal area.
42
What Exactly is Lost?
Early on, the view was this these patients seem
to have a functioning long-term memory as they
can remember things that occurred prior to the
trauma, and they seem to have an intact
working memory as their span, for example, is
relatively normal. What seems to be lost is the
ability to transfer information from working
memory into long-term memory. However, more
recent evidence suggests that such patients are
not completely unable to transfer things to
long-term memory ...
43
Intact Implicit Memories
These patients can learn to stimulus-response
mappings quite well even though they do not
remember learning them mirror drawing,
computer programing, recognition What really
seems to be damaged is their ability to
consciously remember their explicit
memory. Thus, it may not be that information is
not getting in, it may instead be the case that
information cannot be consciously retrieved