Statistical significance . Were the results of the test (experiment 1) statistically significant? How can you tell? What was the p-value? What was the value for the statistical test (e.g., t, F, etc.)? Is the result consistent or inconsistent with the researcher’s hypothesis? How can you tell?
Effect size . Were the statistical test (experiment 1) results accompanied by a measure of effect size? If so, what measure was used and how would you characterize the size of the effect (using Cohen’s rules of thumb)? If no measure of effect size was given, can you suggest one that should have been reported? Can you calculate it yourself from the information that they did provide? Did the authors comment on the size of the effect (either in the Results or Discussion section)?
(
Transcribed Image Text: 570
occurs only when a given percept shows sufficient match with
a specific internal representation. Indeed, various short- and
long-term processes of episodic memory affect object-identifi-
cation time and naming accuracy (e.g., Bentin & McCarthy,
1994; Stuss, Picton, Cerri, Leech, & Stethem, 1992; Verfaellie,
Gabrieli, Vaidya, Croce, & Reminger, 1996).
Following literature suggesting common neural substrates
for memory and perception, we investigated whether inhibi-
tory control exerted on memory representations might affect
later perception of suppressed items. In three experiments, we
used a modified think/no-think paradigm, in which a percep-
tual-identification task replaced the final cued-recall task of
the original paradigm. We hypothesized that if inhibitory
mechanisms exert direct control over perceptual representa-
tions, then successful suppression of object memories would
accompany impaired subsequent identification of the objects.
Experiment I
In our first experiment, we briefly presented object images to
test individuals' ability to perceptually identify memory-sup-
pressed items.
Method
Twenty-five undergraduates with normal or corrected-to-nor-
mal vision and normal color perception participated in
exchange for course credit. The stimuli were 40 critical and 10
filler pairs of a noun and a line drawing of an unrelated famil-
iar object (Snodgrass & Vanderwart, 1980). The word and
drawing in each pair were affectively neutral. Among critical
pairs, 10 pairs each were randomly assigned to think and no-
think conditions, and the remaining 20 pairs were assigned to
the baseline condition. The design of the experiment consisted
of a single factor with three levels (baseline, think, or no-
think) that was manipulated within participants. Experiment 1
consisted of four phases: pretraining object identification,
associative learning, think/no-think training, and posttraining
object identification.
Pretraining object identification. To account for excessive
individual differences, we first measured participants' ability
to identify briefly presented objects. Twenty line drawings of
objects different from the critical stimuli were presented indi-
vidually for 33 ms each in the center of a screen. Presentation
was preceded by a 400-ms fixation period and followed by a
100-ms pattern mask of random line segments. Participants
were asked to write down the names of each object on an
answer sheet. Trials were self-paced.
Associative learning. In the associative-learning phase, the
two items in each stimulus pair were presented concurrently
for 5 s. Pairs were presented individually, with a 600-ms inter-
stimulus interval between each presentation. Participants were
instructed to memorize each association for a later memory
test. After the initial learning cycle, participants were asked to
Kim, Yi
recall corresponding target objects when probed with cue
words. The learning phase was repeated up to four times until
cued-recall accuracy reached 50% at minimum.
Think/no-think training. Each trial of think/no-think training
consisted of a 200-ms fixation cross (either green or red) and a
4-s cue word. In the think condition (indicated by a green fixa-
tion cross), participants were instructed to think of the target
drawing when the cue word appeared. In the no-think condi-
tion (indicated by a red fixation cross) participants were
instructed not to think of the target drawing when the cue word
appeared, thus preventing it from entering their consciousness.
Twenty cue words (10 each for the think and no-think condi-
tions) were presented 12 times in random order. Trials were
separated by 400-ms intertrial intervals. Cue words assigned
to the baseline condition were not presented during the
training.
Posttraining object identification. The procedure for post-
training object identification was identical to the procedure for
the pretraining phase except that 40 critical target drawings
were presented individually either on the left or the right side
of a screen. The location and order of presentation were ran-
domly determined.
Results and discussion
Data from 5 participants whose identification performance
during pretraining was excessively low (accuracy ≤ 10%)
were excluded from the main analysis, which left data from 20
participants for the analysis. The percentage of correctly iden-
tified objects in posttraining in the three conditions was sub-
mitted to a one-way repeated measures analysis of variance
(ANOVA). The effect of condition was significant, F(2, 38) =
3.35, MSE = 164.45, p = .046, n ² = .15 (Fig. 1). Participants
identified significantly fewer objects in the no-think condition
(M = 35%) than in the baseline condition (M= 42.8%), t(19) =
2.55, p = .02, d = 0.85. In contrast, no significant difference
was observed between the percentage of objects identified in
the think condition (M = 45%) and in the baseline condition,
p> .5. These results indicate that suppressing memories of
visual objects impairs subsequent identification of those
objects when they are briefly presented.
Experiment 2
To generalize the results of Experiment 1, we presented noise-
occluded images in the object-identification phase and com-
pared the amounts of information required for correct
identification between conditions.
Method
Twenty-six undergraduates with normal or corrected-to-
normal vision and normal color perception participated for
course credit. In this experiment, we used the same stimuli and
Perceptual Consequences of Memory Suppression
Correctly Identified Objects (%)
60
55
40
35
30
25
20
Baseline
Think
Condition
No-Think
Fig. I. Results from Experiment 1: mean percentage of correctly identified
objects as a function of condition. Error bars represent 95% within-subjects
confidence intervals (Loftus & Masson, 1994).
design as in Experiment 1 but only three of the phases
(associative learning, think/no-think training, and object iden-
tification). The first two phases were identical to those in
Experiment 1. In the object-identification phase, each target
drawing was initially presented with 100% black-and-white
pixel noise. Participants were instructed to reduce the noise
level by pressing a button until they could identify the object
in the drawing. One button press was equivalent to 1% noise
reduction. On reaching the point of object identification, par-
ticipants wrote down the name of the object. Trials were
self-paced.
Results and discussion
The mean maximum percentage of noise allowing for correct
object identification in each of the three conditions was sub-
mitted to a one-way repeated measures ANOVA. This analysis
yielded a significant effect of condition, F(2, 50) = 4.83,
MSE=1.16, p= .012, ₂² .16 (Fig. 2). The maximum noise
level was significantly lower in the no-think condition (M =
74.5%) than in the baseline condition (M = 75.3%), t(25) =
2.31, p= .029, d = 0.59. No difference was found between the
maximum noise level in the think condition (M = 75.4%) and
in the baseline condition, p > .2. These results again demon-
strate impaired identification of memory-suppressed objects
by showing that more perceptual evidence (i.e., noise reduc-
tion) is needed for correct identification.
In a separate supplemental experiment with 9 participants,
we tested whether the .8% lower maximum noise level in the
Maximum Noise (%)
77.0
76.5
76.0
75.5
75.0
74.5
74.0
73.5
73.0
72.5
11
Think
Condition
Baseline
No-Think
571
Fig. 2. Results from Experiment 2: mean maximum percentage of noise
allowing for objects to be correctly identified as a function of condition.
Error bars represent 95% within-subjects confidence intervals (Loftus &
Masson, 1994).
no-think condition than in the baseline condition was func-
tionally meaningful in object identification. The procedure
was identical to the one used in the pretraining phase of Exper-
iment 1, except that 40 critical target drawings, covered with
either 75.3% or 74.5% noise, were presented for 400 ms, each
followed by a 600-ms mask. Paired-samples t ests revealed
that target objects covered with 74.5% noise were significantly
better identified than those covered with 75.3% noise (Ms
42.2% vs. 31.7%, respectively), t(8) = 2.41, p = .042, d = 2.52
(Fig. 3). This result confirms that the 0.8% noise reduction
found in Experiment 2 bears significance in improving per-
ceptual-identification performance.
Experiment 3
Impaired object identification resulting from suppression
training could be due to inhibition of perceptual representa-
tions, inhibition of conceptual representations, or both. In
Experiment 3, we examined the relative contribution of these
factors on impaired identification by presenting mirror-
reversed images of target objects in the object-identification
phase. We predicted that if the observed impairment were
mainly due to the inhibition of conceptual representations,
participants would show a similar amount of identification
impairment as in Experiment 2, irrespective of slight changes
in perceptual information. However, if the inhibition of per-
ceptual representations was the main factor contributing to the
impaired identification, the effects of memory suppression on
object identification should disappear in Experiment 3.