WHEN THE PROBE IS WILLING BUT THE MIND IS WEAK This research was supported by an NSERC operating...
-
Upload
chrystal-carter -
Category
Documents
-
view
215 -
download
1
Transcript of WHEN THE PROBE IS WILLING BUT THE MIND IS WEAK This research was supported by an NSERC operating...
WHEN THE PROBE IS WILLING BUT THE MIND IS WEAKWHEN THE PROBE IS WILLING BUT THE MIND IS WEAK
This research was supported by an NSERC operating grant
For additional information please contact Marty Niewiadomski at [email protected]
Marty Niewiadomski University of Toronto
at Scarborough
Marty Niewiadomski University of Toronto
at Scarborough
Steve Joordens University of Toronto
at Scarborough
Steve Joordens University of Toronto
at Scarborough
Bill Hockley Wilfrid Laurier
University
Bill Hockley Wilfrid Laurier
University
A New Recognition Paradigm:
THE WHAT• Partial-item recognitionTHE HOW
•Study List– Present a list of words one at a time.
•Partial Item Recognition Test– 48 “old” and 48 “new” test probes were
presented one at a time.– All of the items had two letters removed.– All partial items could only be completed with
one solution.– Subjects were asked to first generate the
missing letters to complete the word, and then make a new/remember/know recognition decision for the completion
THE WHY• Reason # 1 – The paradoxical frequency effect
in recall and recognition.• In a recognition test, participants typically
make more hits and fewer false alarms to low-frequency words compared to high frequency words (mirror effect). In a recall test, however, a general memory advantage favours high frequency items, as they tend to be recalled more often relative to low frequency words.
• Experiment 1 deals directly with this paradox.• Reason # 2 – Beyond good encoding and
environmental support – the role of context in recognition memory.
• Experiments 2-5 explore the roles of environmental support and context within the mirror effect.
• In all the Experiments, the partial-item recognition task helps to disambiguate the issues.
A New Recognition Paradigm:
THE WHAT• Partial-item recognitionTHE HOW
•Study List– Present a list of words one at a time.
•Partial Item Recognition Test– 48 “old” and 48 “new” test probes were
presented one at a time.– All of the items had two letters removed.– All partial items could only be completed with
one solution.– Subjects were asked to first generate the
missing letters to complete the word, and then make a new/remember/know recognition decision for the completion
THE WHY• Reason # 1 – The paradoxical frequency effect
in recall and recognition.• In a recognition test, participants typically
make more hits and fewer false alarms to low-frequency words compared to high frequency words (mirror effect). In a recall test, however, a general memory advantage favours high frequency items, as they tend to be recalled more often relative to low frequency words.
• Experiment 1 deals directly with this paradox.• Reason # 2 – Beyond good encoding and
environmental support – the role of context in recognition memory.
• Experiments 2-5 explore the roles of environmental support and context within the mirror effect.
• In all the Experiments, the partial-item recognition task helps to disambiguate the issues.
– If subjects could not generate the appropriate letters, or made a mistake, the computer would fill in the blanks or correct them.
– If subjects could not generate the appropriate letters, or made a mistake, the computer would fill in the blanks or correct them.
ME__RY
MEM_RY
MEMORY
MEMORY
1=Remember OLD 2=Think OLD 3=NEW
TEST LISTPARTIAL ITEM RECOGNITION
EXPERIMENT 1 – FREQUENCY PARADOX
METHOD• Study list
– 60 words were presented one at a time.– Words were high frequency (mean = 143) or
low frequency (mean = 6.9), blocked by frequency (blocks were counterbalanced and no block effect was observed in any experiments).
– Half of the old and the new items were high frequency and half were low frequency. Presentation of old/new probes as well as word frequency was randomized at test.
RESULTSGeneration• A generation advantage was observed for
high-frequency items relative to low frequency items
• This advantage was present for both “old” and “new” items.
Recognition• A typical mirror pattern was observed
between the new and old items. Furthermore, the recollective process usually associated with “remember” responses appears to be responsible for the “old” portion of the effect.
EXPERIMENT 1 – FREQUENCY PARADOX
METHOD• Study list
– 60 words were presented one at a time.– Words were high frequency (mean = 143) or
low frequency (mean = 6.9), blocked by frequency (blocks were counterbalanced and no block effect was observed in any experiments).
– Half of the old and the new items were high frequency and half were low frequency. Presentation of old/new probes as well as word frequency was randomized at test.
RESULTSGeneration• A generation advantage was observed for
high-frequency items relative to low frequency items
• This advantage was present for both “old” and “new” items.
Recognition• A typical mirror pattern was observed
between the new and old items. Furthermore, the recollective process usually associated with “remember” responses appears to be responsible for the “old” portion of the effect.
0.25
0.480.43
0.63
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Pro
po
rtio
n G
ener
ated
New Old
Low Freq
High Freq
0.260.32
0.46 0.46
0.820.74
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Pro
po
rtio
n o
f O
ld R
es
po
ns
es
New Know Remember
Low Freq
High Freqov
eral
l
• This general pattern of results can be seen regardless of whether the items were generated or not.
EXPERIMENTS 2-5: ENVIRONMENTAL SUPPORT AND THE MIRROR EFFECT IN
RECOGNITION
• Two of the factors that influence memory are encoding and environmental support at test; the partial-item recognition task allows easy manipulation of the latter one. Simply removing letters from items at test decreases the amount of environmental support.
• These experiments used the partial-item recognition paradigm in three different settings: either two letters were removed, one letter was removed or no letters were removed (standard yes/no recognition).
• These conditions were tested in a between-subjects design (Exp 2-4) and in a within-subjects design (Exp. 5).
• This general pattern of results can be seen regardless of whether the items were generated or not.
EXPERIMENTS 2-5: ENVIRONMENTAL SUPPORT AND THE MIRROR EFFECT IN
RECOGNITION
• Two of the factors that influence memory are encoding and environmental support at test; the partial-item recognition task allows easy manipulation of the latter one. Simply removing letters from items at test decreases the amount of environmental support.
• These experiments used the partial-item recognition paradigm in three different settings: either two letters were removed, one letter was removed or no letters were removed (standard yes/no recognition).
• These conditions were tested in a between-subjects design (Exp 2-4) and in a within-subjects design (Exp. 5).
TEST LISTINTACT
SI_TER
NEW OLD
NE_ULA
NEW OLD
CO_KIE
NEW OLD
RE_ALL
NEW OLD
RE_SON
NEW OLD
PEOPLE
NEW OLD
SI_TER
NEW OLD
NE_ULA
NEW OLD
CO_KIE
NEW OLD
RE_ALL
NEW OLD
RE_SON
NEW OLD
PE_PLE
NEW OLD
TEST LIST1 LETTER REMOVED
TEST LIST2 LETTERS REMOVED
SI_TER
NEW OLD
NE_ULA
NEW OLD
CO_KIE
NEW OLD
RE_ALL
NEW OLD
RE_SON
NEW OLD
PE__LE
NEW OLD
0.230.28
0.41 0.40
0.78
0.67
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Pro
po
rtio
n o
f O
ld R
es
po
ns
es
New Know Remember
Low Freq
High Freqnon
-
gener
ated
0.36 0.380.42
0.45
0.86
0.76
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
Pro
po
rtio
n o
f O
ld R
es
po
ns
es
New Know Remember
Low Freq
High Freqgener
ate
d
RESULTS
Between Subjects
• A typical mirror effect was observed in the intact condition (no letters removed).
• The “old” portion of the mirror effect was significantly attenuated in both partial-item conditions.
Within Subjects
• No memory advantage was observed in the “old” portion of the effect, regardless whether the item had two, one, or no letters removed.
CONCLUSIONS
•Experiment 1 shows a HF advantage for generation and a LF advantage for recognition, all within one experiment.
•Results of Experiments 2-5, show that good encoding and environmental support are not always sufficient to support recollection.
•Subjects may adopt a common strategy in dealing with partial recognition probes that results in decreased recollection.
•Recognition results are consistent with 2-Factor accounts of the mirror effect (e.g. Joordens & Hockley, 2000).
RESULTS
Between Subjects
• A typical mirror effect was observed in the intact condition (no letters removed).
• The “old” portion of the mirror effect was significantly attenuated in both partial-item conditions.
Within Subjects
• No memory advantage was observed in the “old” portion of the effect, regardless whether the item had two, one, or no letters removed.
CONCLUSIONS
•Experiment 1 shows a HF advantage for generation and a LF advantage for recognition, all within one experiment.
•Results of Experiments 2-5, show that good encoding and environmental support are not always sufficient to support recollection.
•Subjects may adopt a common strategy in dealing with partial recognition probes that results in decreased recollection.
•Recognition results are consistent with 2-Factor accounts of the mirror effect (e.g. Joordens & Hockley, 2000).
0.24
0.35
0.72 0.73
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Pro
po
rtio
n o
f O
ld R
esp
on
ses
New Old
Low Freq
High Freq
0.31
0.40
0.630.60
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Pro
po
rtio
n o
f O
ld R
esp
on
ses
New Old
Low Freq
High Freq
0.12
0.27
0.78
0.68
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Pro
po
rtio
n o
f O
ld R
esp
on
ses
New Old
Low Freq
High Freq
0.13
0.28
0.66
0.76
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Pro
po
rtio
n o
f O
ld R
esp
on
ses
New Old
Low Freq
High Freq
0.21
0.32
0.69 0.72
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Pro
po
rtio
n o
f O
ld R
esp
on
ses
New Old
Low Freq
High Freq
0.28
0.40
0.570.64
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
Pro
po
rtio
n o
f O
ld R
esp
on
ses
New Old
Low Freq
High Freq
betw
een
Ss
–
inta
ct
betw
een S
s –
1 lett
er
rem
oved
betw
een S
s –
2 lett
ers
rem
oved
with
in S
s – 1
l
ette
r
rem
oved
with
in S
s – 2
lett
ers
rem
oved
with
in
Ss
–
inta
ct