Introduction and Methodology
• How to obtain empirical evidence for distinct systems
participating together in a global function?
System A
System B
• Establishing a dissociation
• if two different systems participate together in a global
function, then they may be distinguished empirically by
showing that each system reacts to a different set of variables
• this can be done by constructing a task for each particular
system
Introduction and Methodology
• Variables are like different measurement instruments or
chemical agents that cause a specific reaction in a particular
cognitive system
System B
System A
Variable 1 affects
system A
Variable 1 does not
affect system B
(single dissociation)
Introduction and Methodology
System A
Variable 1 affects
system A
System A could be a memory
system that temporarily holds
a few items (e.g., telephone
number) just as long as
one thinks about it
System B
Variable 1 does not
affect system B
System B could be a memory
system with a very large
capacity and holds information
over a long duration
(e.g., word meaning)
Introduction and Methodology
• What variable might specifically affect performance of
System A?
• acoustic (auditory) similarity
• e.g., MAN, MAD, CAP, CAN, MAP
• How do we prove that System A is affected by acoustic
similarity?
• show that performance is worse on an appropriate task for
words that are acoustically similar
• compared to words that are not acoustically similar, but
matched in all other possible respects to the similar set
Similar: MAN, MAD, CAP, CAN, MAP
Control: PEN, RIG, DAY, BAR, CUP
Introduction and Methodology
• What task might we use to engage System A?
• remembering a short list of items in serial order
• This is the task that cognitive psychologists have developed
to examine the effect of acoustic similarity on this memory
system
• Listen to the following lists of 5 words
• after each list, write down the words in exactly the order in
which they occurred.
List 1: MAN, MAT, CAP, CAD, CAT
List 2: COW, BUN, RIG, FEW, DAY
Introduction and Methodology
• Might difficulty with acoustically similar lists be due to the fact
that it is harder to tell the words apart as auditory signals?
• will the difficulty disappear if the words are presented visually
to avoid this problem?
• We can test this possibility using a visual presentation of the
list
CAP
MAP
MAN
CAB
MAT
List 3: MAT, CAB, CAP, MAN, MAP
Introduction and Methodology
• Here are the results from the original Baddeley (1966) study
Introduction and Methodology
• To show that System A can be dissociated from System B, what
might be the next step?
• Show that System B would not be affected by acoustic
similarity
Introduction and Methodology
• We need a task that will specifically engage System B
• what would this task be like?
System A
System A could be a memory
system that temporarily holds
a few items (e.g., telephone
number) just as long as
one thinks about it
System B
System B could be a memory
system with a very large
capacity and holds information
over a long duration
(e.g., word meaning)
Introduction and Methodology
• Answer: A long list of words (e.g., 10) that has to be learned
to some reasonable degree of accuracy
System A
System B
Task B: Remember this list…
PIT, FEW, COW, PEN, CUP, BAR, DAY, HOT, RIG, BUN
• We face a problem if we try the task this way… what is it?
• Answer: System A will contribute to performance of Task B
because it can hold a short list of items for a brief time
Introduction and Methodology
• How can we solve this problem?
System A
System B
Task B: Remember this list…
PIT, FEW, COW, PEN, CUP, BAR, DAY, HOT, RIG, BUN
• Present the long list, but before asking for recall, add a
preliminary task that will engage System A for some time
• what would this preliminary task be like?
Introduction and Methodology
• In summary, Task B will consist of 3 steps
• present the long list of words
• present an intervening System A task (e.g., present and
require serial order report of a set of 6 digits)
• recall of the long list of words
• Task B performance should reflect System B without the
influence of System A
Introduction and Methodology
• But there is another issue to consider
• imagine what it would be like to hear a list of 10 words, then
hear and recall in order a list of 6 digits, then finally recall the
10 words
• What would we initially see when we ask for recall of the 10
words?
• recall will not be very good
• it may not be possible to see any difference between an
acoustically related list and a control list
• What should we do?
• give subjects multiple opportunities to learn and to
attempt to recall the list
• examine performance after a number of learning attempts
Introduction and Methodology
• So here is what we must do
1. List of 10 words
Repeat this cycle until a
2. List of six digits
reasonable level of accuracy
3. Recall digits in order
is reached
4. Recall words
]
• We use two lists of words (two conditions)
List 1: acoustically similar
List 2: control list for list 1
Introduction and Methodology
• One last problem
• in what way should we have subjects report the words?
• suppose subjects were told to recall the words in any order
• e.g., MAP, PAT, RAT, LAP, SAT, HAT, CAT, CAP, SAP, BAT
• what strategy might a subject use that would corrupt our
intention to assess System B?
• Subjects might use the category (“words ending in AT or AP”)
to generate relevant members without trying to recall
which ones were actually on the list
• How can we ensure that subjects recall the specific items that
were on the list?
• require report in the original presentation order
Introduction and Methodology
• Now let’s try a demonstration of Task B in its full form
• first you will see a list of words that you are to learn for later
recall
• then you will hear a series of six digits
• next you are to report the six digits in their original order
• finally, try to report as many of the words from the list in their
original order
Introduction and Methodology
GAP
MAP
CAP
NAP
BAT
CAT
TAP
HAT
VAT
FAT
Now report the word list
715043
GAP CAT NAP CAP BAT VAT TAP FAT MAP HAT
Introduction and Methodology
• Here is the result from the Baddeley (1966) study
• note lack of effect on trial 4
• why might there be some
effect on early trials?
Introduction and Methodology
• Now we have a single dissociation between Systems A and B
based on a manipulation of acoustic similarity
Effect on System A
No effect on System B
Introduction and Methodology
• Now we will try to establish a double dissociation by identifying
a manipulation that will affect System B but not System A
• what kind of information does System A not use in
registering words in memory but that System B is likely to
use?
the meaning of words
System A
System A could be a memory
system that temporarily holds
a few items (e.g., telephone
number) just as long as
one thinks about it
System B
System B could be a memory
system with a very large
capacity and holds information
over a long duration
(e.g., word meaning)
Introduction and Methodology
• We can repeat the previous experiment using words that are
semantically confusable, but not similar acoustically
• a short list for System A
• a long list for System B
• Again, for both tasks subjects are required to report items in
their original order
Introduction and Methodology
• Listen to the following list of 5 words
• after the list, write down the words in exactly the order in
which they occurred.
List 1: WIDE, LARGE, HIGH, BIG, TALL
• Now we again have to compare performance on the confusable
list to performance on a suitable control list
List 2: SAFE, DEEP, OLD, GOOD, LATE
Introduction and Methodology
• These are the full results from Baddeley (1966)
System A
System B
Long lists – 4th trial
Acoustic Sim.
Semantic Sim.
Difference (%)
70
60
50
40
30
20
10
0
Task A
Task B
Acoustic Sim.
Semantic Sim.
Difference (%)
70
60
50
40
30
20
10
0
Task A
Task B
Acoustic Sim.
Semantic Sim.
Difference (%)
70
60
50
40
30
20
10
0
Task A
Task B
Acoustic Sim.
Semantic Sim.
Difference (%)
70
60
50
40
30
20
10
0
Task A
Task B
Acoustic Sim.
Semantic Sim.
Difference (%)
70
60
50
40
30
20
10
0
Task A
Task B
Introduction and Methodology
• Double dissociation methodology with neurological cases
• imagine System A is badly damaged, but not System B
Task A: repeat a short list in order
Task B: learn a long list in order
• What effects are expected on these tasks?
Task A impaired; Task B normal
Introduction and Methodology
• Double dissociation methodology
• imagine System B is badly damaged, but not System A
Task A: repeat a short list in order
Task B: learn a long list in order
• What effects are expected on these tasks?
Task A normal; Task B impaired
Introduction and Methodology
• Example of damage to System A but not to System B
• Warrington & Shallice (1969)
• case study: K.F. injured in a motorcycle accident at age 17
• skull fracture on left side
• surgery to relieve swelling and bleeding
• major region affected: left parietal lobe
• research conducted when K.F. was 28 years old
Introduction and Methodology
• Effects on K.F.’s abilities
• well preserved understanding of spoken language
• impaired verbal expression (halting speech, word-finding
difficulty)
• slow but accurate reading
• impaired spelling and writing
• Serious impairment of ability to repeat short list of verbal items
• no problem if presented with a single number, letter, or word
• marked decline in performance for lists of two or more items
Introduction and Methodology
Introduction and Methodology
• Normal performance when learning a list of 10 words over a
series of trials (Task B)
• K.F. needed 7 learning trials for a list of 10 words
• control subjects needed an average of 9 trials
Introduction and Methodology
• An example of a similar case
• 53-year old woman (R) with a left parietal stroke
• able to maintain normal conversations, remember facts and
events
• In the following sound segment, this patient is being ask to
repeat pairs of nonsense words
Examiner: HOUN GEN
DOUM HAR
SEV
HUL
Introduction and Methodology
• An example of the reverse dissociation
• what might be the characteristics of a person with this
impairment?
• easily repeat back a telephone number
• serious difficulty memorizing a 10-item shopping list
Introduction and Methodology
• The case of H.M.
• underwent surgery to relieve severe epileptic seizures
Introduction and Methodology
• Testing for System A function: digit span task
2 9714723 6
• Normal span is 6 or 7 digits
• suppose a persons' span is 6 digits--what happens if we add
one more item to the list?
4371825
• Suppose you make a mistake, so we try the same list again
4371825
Introduction and Methodology
• If you get the list right, we give you another list that is one item
longer in length--a total of 8 items
• if you fail on the 8-item list, we try that list again
• We provide 25 attempts at the same list before giving up
• What is the maximum list length you can learn in this way?
• Extended digit span
• What memory system does it require?
• both System A and System B
Introduction and Methodology
• Suppose System B is badly impaired--what would digit span
performance be like relative to extended digit span for such a
patient?
• very much the same
Introduction and Methodology
• Summary of the full double dissociation that supports the
existence of two different systems (A and B)
• Construct a plot of effects showing the full double dissociation
KF vs. Controls
HM vs. Controls
4
3
2
1
0
Task A
Task B
KF vs. Controls
HM vs. Controls
4
3
2
1
0
Task A
Task B
KF vs. Controls
HM vs. Controls
4
3
2
1
0
Task A
Task B
KF vs. Controls
HM vs. Controls
4
3
2
1
0
Task A
Task B
KF vs. Controls
HM vs. Controls
4
3
2
1
0
Task A
Task B
KF vs. Controls
HM vs. Controls
4
3
2
1
0
Task A
Task B
Introduction and Methodology
9. In an experiment, Rogers (1974) had observers learn the
names of eight schematic faces (the faces and names are shown
in the figure below). Each face-name pair can be considered an
"individual."
After the learning experience, the observers performed the
following task: On each trial, the observer was presented with a
name printed below a face and was required to indicate with a key
press whether the name correctly matched the face (left key press
"yes", right key press = "no").
On half the trials, the name was the one associated with the face
during learning, whereas on the remaining trials, mixed in
randomly with the positive trials, the name did not correctly match
the face but rather matched a different one of the faces. For "no"
trials, faces and names were selected so that they differed by
either one or two features (e.g., the faces of DAPA and DOPO
differ only in respect to their mouth, but the faces of DAPA and
DALA differ in respect to the mouth and the shape of the head;
the names DOPO and DOLO differ by one feature [letter], while
the names DOPO and NOLO differ by two features [letters]).
So, for example, if presented with the face of DOPO and the
name "DOLO" on a "no" trial, the presented face differs from the
face corresponding to the presented name in one feature (the
eyes––see DOPO and DOLO in the figure). The presented name
differs from the name corresponding to the presented face in one
feature as well ("L" versus "P").
The figure below indicates the correct average response times
(RT's) on these "no" trials. Remember that on each "no" trial, two
different learned "individuals" (face-name pairs) are
represented––one by its name and one by its face. The dotted
line indicates RT when the presented name differs from the
presented face's actual name by two features (e.g., DOPO differs
from NOLO by two letters) and the solid line indicates RT on "no"
trials when the presented name differs from the presented face's
actual name by one feature (e.g., DOPO versus DOLO).
Remember that the task is always to indicate whether the name
matches a face presented on the screen at the same time.
(a) What is the magnitude of the effect on RT of one versus two
features that differ between faces when names differ by only one
feature?
(b) What is the magnitude of the effect on RT of one versus two
features that differ between faces when names differ by two
features?
(c) Consider the values obtained and answer "yes" or "no" to the
following question: Do these values suggest that the observers
are relying on a visual representation of two faces when they
attempt to judge whether a name and a face correspond to the
same individual?
(d) Now calculate the average response time when the names of
the faces differ by one letter, regardless of whether the actual
faces differ by one or two face features.
(e) Calculate the average response time when the names of the
faces differ by two letters, regardless of whether the actual faces
differ by one or two face features.
(f) Do the results suggest that the observers are using a name
representation of two faces when they attempt to judge whether a
name and a face correspond to the same individual? Answer
"yes" or "no".
Now look at the figure below, which shows the results of a variant
on this experiment. In the left section of the figure, we see the RT
on "no" trials when the face was presented first and the name was
presented 2 seconds later. Results in the right section of the
figure are from trials on which the name is presented first, and the
face occurs 2 seconds later.
(g) When the face occurs first, is RT longer when the individuals
differ by only one face feature compared to two face features?
Simply indicate "yes" or "no".
(h) When the face occurs first, is RT longer when the names of
the individuals differ by only one feature compared to two
features? Simply indicate "yes" or "no".
(i) According to the figure, when the face occurs first and there is
a 2-second delay until the name occurs, which of the following
sources of information is used to judge that the name DALA, say,
does not match the face? Select one option.
(1) Visual representations of the faces only.
(2) Representations of the names only.
(3) Both visual representations of the faces & representations of the names.
(j) According to the figure, when the name occurs first, which
representation does the observer rely on to carry out the task?
Select one option.
(1) Visual representations of the faces only.
(2) Representations of the names only.
(3) Both visual representations of the faces and representations of the
names.
(k) Plot a bar graph using the axes below to show that the data in
the figure above constitute a double dissociation between name
similarity and face similarity for the two tasks–Task 1: faces
followed by names; Task 2: names followed by faces. Label the
bars in your graph appropriately.