Full text of DOI:10.3758/BF03194938

Memory & Cognition
002, 30 (3), 372-379
2
Mental simulation inflates performance estimates
for physical abilities
JOSHUA D. LANDAU
York College of Pennsylvania, York, Pennsylvania
TERRY M. LIBKUMAN
Central Michigan University, Mount Pleasant, Michigan
and
JONATHON C. WILDMAN, JR.
York College of Pennsylvania, York, Pennsylvania
In five experiments, we examined how mental simulation of physical activities affected estimates of
one’s ability to perform the same activities.In Experiment 1, participants who simulated lifting a heavy
object estimated that they could lift more weight than did participants who did not perform the simula-
tion. In Experiment 2A, the frequency with which participants performed the simulation exercises was
manipulated. In Experiments 2B and 2C, we manipulatedthe amount of weight that people simulatedlift-
ing in order to address potential alternativeexplanations of the inflation effect. In Experiment 3, mental
simulations were manipulated within subjects. In all the experiments, the simulated events showed in-
flatedestimates,as compared with nonsimulated events.These resultswere interpretedin the context of
the misattribution-of-familiarityaccount of imagination inflation.
The psychologicalliteratureis replete with examples of (e.g., TREE–DESK). Following each pair, people estimate
situationsin which people possess mistaken beliefs about their confidence that they will be able to remember the
reality. For example, people will believe that they have target word (DESK) when completing a cued-recall test
generated a novel idea when they have inadvertently pla- later in the experimental session. When the JOL immedi-
giarized another person ’s work, because they fail to rec- ately follows learning the word pair, people ’s judgments
ognize that someone else authored the idea (e.g., Landau are highly inaccuratefor predictingfuture recall (i.e., high
&
Marsh, 1997; Marsh & Landau, 1995). In other situa- confidence does not result in accurate recall of the target
tions, people overestimate their own contributions to a words). Inadvertent plagiarism, egocentric bias, and inac-
joint product. Ross and Sicoly (1979) found that when curate JOLs are just three experimentalfindingsthat high-
married couples separately estimated their degree of re- light situationsin which people ’s beliefs and reality are not
sponsibilityfor several activities(e.g., causing arguments, synchronous.
taking out the garbage, etc.), each couple ’s responses re-
When people engage in mental simulations, this seems
vealed an egocentricbias indicatingthat one or both of the to exaggerate the disparity between belief and reality.
spouses overestimated how much they contributed to the One of the more potent demonstrations of this phenome-
majority of the activities. Investigations of judgments of non is imagination inflation (Garry, Manning, Loftus, &
learning (JOL) indicate that people are often inaccurate in Sherman, 1996; Goff & Roediger, 1998; Heaps & Nash,
their predictions of future cognitive performance (e.g., 1999; Hyman & Billings, 1998; Paddock et al., 1998). In
Dunlosky & Nelson, 1994; Nelson & Dunlosky, 1991). In one such experiment, Garry et al. (1996) had people rate
the JOL paradigm, people study a list of paired associates the likelihoodthat they had experiencedcertain childhood
events (e.g., gettingstuck in a tree, finding a $10 bill, etc.).
People returned to the laboratory 2 weeks later and men-
tally simulated some of the events from the first week.
Experiment 2 was presented at the 1999 meeting of the Eastern Psy-
Afterward, people completed the rating task a second
chologicalAssociation. Thanksare expressed to Chad C. Cooper, who as-
time. The rated likelihoodof the simulatedevents that were
sisted in the data collection for Experiment 2, and Richard L. Marsh and
originally assigned low ratings increased on the second
Andrew Leynes, who providedinsightfulcomments on an earlier draft of
the manuscript. Address correspondence to J. D. Landau, Department of questionnaire,indicatingthat mental simulationaltered the
Behavioral Sciences, York Collegeof Pennsylvania,York,PA 17405-7199 participant ’s memories for childhoodexperiences. In con-
or to T. M. Libkuman, Department of Psychology, Central Michigan Uni-
trast, the ratings of the nonsimulatedevents remained rel-
versity, Mt. Pleasant, MI 48859 (e-mail: jlandau@ycp.edu or terry.m.
atively unchanged. These results indicate that mentally
libkuman@cmich.edu).
simulating an event when it has not occurred can have
—
Accepted by previous editorial team deleterious effects on accurate recollections of the past.
Copyright 2002 Psychonomic Society, Inc.
372

MENTAL SIMULATION
373
To date, imagination inflation research has been lim- ence performance predictions. It is also possible that peo-
ited to investigations of how mental simulations affect ple will discountthe information from the mental simula-
memory for past events (i.e., people judge the likelihood tion because they are aware of how much weight they are
that events have taken place sometime in the past). The capable of lifting.
question of how mental simulation affects predictions of
Method
future performance for physical abilities has not been ad-
dressed. This issue is important because many people use
mental simulation to prepare themselves for future cog-
nitive tasks. Taylor and her colleagues(e.g., Pham & Tay-
lor, 1999;Taylor,Pham, Rivkin,& Armor, 1998)suggested
that mental simulation improves performance by permit- condition engaged in a mental simulation exercise in which the fol-
ting people to envisionalternativepossibilities and elim- lowing instructions were read:
Participants. Eighty-seven undergraduates were randomly as-
signed to either the no-simulation (n 5 44) or the simulation (n 5
4
5
3) condition. Testing was conducted in small groups ranging from
to 10 participants.
Materials and Procedure. The participants in the simulation
inate potentialerrors. Mental simulation gives people the
opportunity to gauge whether they are prepared to actu-
allyperform each part of the task. In thissense, mentalsim-
ulation provides people with a way of estimating their
current level of ability.
The main questionthat we examined is whether a brief
mental simulation exercise would have a measurable im-
pact on prospective estimates for physical abilities.
Rather than focus on predictionsof future cognitiveabil-
Close your eyes and imagine that you are helping a friend move and
that it is your responsibility to move the refrigerator. Mentally picture
yourself reaching around the appliance and gently lifting up and hold-
ing it there for a few seconds. Now, ease the refrigerator back to the
ground and move away from it.
After completing the simulation exercise, which lasted approxi-
mately 15 sec, the participants reported the amount of weight they
could benchpress. The participants in the no-simulation condition
simply estimated how much weight they could benchpress.
ities, we examined how mental simulations would influ- Results and Discussion
ence predictions of physical performance. In the experi-
Unless specified otherwise, the Type I error rate was set
ments that follow, participantsmentallysimulatedphysical at .05 for all analyses reported in this paper. The purpose
activities and then estimated their proficiency at the ac- of this experiment was to determine whether a brief men-
tivities. In Experiments 1–2C, the participants engaged tal simulation exercise would have a measurable impact
in a mental simulation exercise in which they imagined on performance estimates. As was predicted, those par-
lifting various amountsof weights and then estimatedhow ticipants who mentally simulated lifting a heavy object
much weight they could lift. In Experiment 1, half of the reported being able to lift more weight (M 5 114 pounds,
participants mentally simulated an event before making SE 5 11.3) than did participantsin the no-simulationcon-
1
the estimation, and the other half did not. In the remain- dition[M 5 88 pounds,SE 5 8.2; F(1,84)5 3.4, p , .05].
ing experiments,several relevantvariableswere examined We believethat the weight estimates increased because the
to determine their influence on the weight estimates. In mental simulation exercise made it easier for people to
Experiment2A, we manipulatedthe frequency with which believe that they could lift more weight. Goff and Roedi-
people performed the mental simulationexercise to ascer- ger (1998) speculated that imagination inflation for ret-
tain whether it would affect their weight estimates (e.g., rospective memories results from a confluence of factors.
Goff & Roediger, 1998). In Experiments 2B and 2C, we They argued that simulating an event increases its famil-
varied the amount of weight that the participantsimagined iarity, which makes it easier to bring the event to mind.
lifting. Finally,to increase generality, in Experiment 3 we Later, when makingjudgmentsabouttheevent,peoplemis-
used a within-subjects design in which participants sim- take the fluency with which the idea comes to mind for
ulated several different physicaleventsand then rated their evidence that the event actually occurred in the past. Peo-
abilities at these events.
ple misinterpret the increased familiarity as an indication
of a real event, and not of a simulatedevent (e.g., Johnson,
1988; Johnson, Raye, Hasher, & Chromiak, 1979). There
is a substantial body of evidence suggesting that people
EXPERIMENT 1
We hypothesized that mental simulation of an event do not sufficiently monitor the source of the familiarity
would inflate performance predictionestimatesfor a phys- when making judgments such as confidence ratings and
ical ability. One group of participantsmentally simulated task difficulty (Jacoby & Kelley, 1987; Johnson, Hash-
lifting a heavy object (a refrigerator), whereas a second troudi, & Lindsay, 1993; Kelley & Jacoby, 1996;Kelley &
group did not perform the mental simulation. The partic- Lindsay, 1993; Marsh, Landau, & Hicks, 1997). Conse-
ipants then estimated how much weight they could lift. quently, when judgingthe occurrence of a simulatedevent,
According to our literature review (e.g., Goff & Roediger, people confuse the source of the familiarity and incor-
1
998; Taylor et al., 1998), if mental simulation affects rectly claim that the event was actuallyexperienced,when,
predictionsabout future performance, people in the men- in fact, it was not. In a similar fashion, when people made
tal simulationconditionshould provide higher weight es- their weight estimates after the refrigerator simulationex-
timates than do participants in the control condition. Al- ercise, the details of the mental simulation exercise came
ternatively, our mentalsimulationexercise,which was brief to mind easily. Rather than discounting this information
and lacking in detail, might not be potent enough to influ- as the residue from the simulation exercise, people used

3
74
LANDAU, LIBKUMAN, AND WILDMAN
it as a basis for their judgments. Consequently, they re- imagery questionnaireand thehealthsurvey shouldreduce
ported inflated weight estimates.
the likelihoodthat the people would intentionally inflate
Although the misattribution account is plausible, there their weight estimates. In order for the people to intention-
are several alternative explanations to consider. First, it ally inflate their weight estimates, they would first have to
is possible that the participants assigned to the simula- discover the relationship and then decide to add weight
tion condition could actually lift more weight than those to their estimates as a way to appease the experimenter.
assigned to the no-simulationcondition.This explanation
If inflation of weight estimates and imagination infla-
is unlikely, because we randomly assigned participants tion (Garry et al., 1996; Goff & Roediger, 1998) are pro-
to the two conditions. Second, it is possible that the peo- duced by the same mechanism, manipulations that influ-
ple in the simulationconditionprovided higher weight es- ence imaginationinflationshould have similar effects on
timates because of demand characteristics. That is, they weight estimates. In their investigation of imagination
believed that the experimenter desired a higher estimate, inflation, Goff and Roediger had participants either per-
and they inflated their estimates in an effort to help the form or imagine simple activities(e.g., breaking a tooth-
experimenter. Previous literature, however, suggests that pick in your mouth, tuggingyour earlobe, etc.) one, three,
demand characteristics are not a viable explanation.Gre- or five times. The participantsthen decided whether they
gory, Cialdini, and Carpenter (1982) reported that im- had actually performed a list of activities. Some of these
agery inflation was not due to demand characteristics. In activitieshad been performed earlier, some had only been
their experiment,people imaginedbeingarrested for petty imagined, and some were new activitiesthat were neither
theft and shoplifting. People in a control group read neu- performed nor imagined. As the number of imagination
tral scenarios that were not crime related. The researchers trials increased, people came to believe they had actually
contacted the people later that evening as part of a seem- performed an activity, when, in fact, they had not. On the
ingly unrelated telemarketing survey. Each person an- basis of this finding, we manipulatedthe number of times
swered a series of questions, includingan estimate of how that people mentally simulated lifting 200 pounds (0, 5,
likely it was that they would be arrested for different or 10 simulationtrials)in Experiment2A. We hypothesized
crimes. People in the imaginationconditionthought that it that increasing the number of simulation trials would in-
was more likely that they could be arrested for petty theft crease weight estimates. Moreover, if demand character-
or shoplifting.Importantly, none of thepeopleclaimed that istics were operating in this experiment, the people per-
they were aware of the relationship between the experi- forming 10 simulationswould have to intentionallydecide
mental session and the survey.
to add more weight than would the people performing 5
Gregory et al.’s (1982) results indicate that people will simulations.
inflatetheirestimates even when the potentialcontribution
The primary motivation for Experiment 2B was to de-
of demand characteristics is small. The previous literature termine whether the magnitude of the weight mentioned
notwithstanding,we designed the following experiments, in the mental simulationexercise would affect prospective
in part, to investigate the demand characteristic explana- weight estimates. Other researchers have found that the
tion more fully. In Experiments 2A, 2B, and 2C we used a details of the mental simulation exercise influence perfor-
disguised paradigm that was conceptually similar to the mance (e.g., Intons-Peterson & Roskos-Ewoldsen, 1989;
one used by Gregory et al. Disguising the weight estimate Taylor et al., 1998). For example, Intons-Peterson and
as part of a separate, unrelated study should decrease the Roskos-Ewoldsen asked people to mentally travel speci-
probabilitythat peopleintentionallyinflate theirweight es- fied routes and to press a button when they arrived at the
timates in an effort to help the experimenter.
destinationpoint.During this mental trip, peopleimagined
transportingobjectsof increasing weight (e.g., a balloon,
a ball, and a cannonball). As the weight of the objects in-
creased, so did the mental transport times. Thus, the ac-
EXPERIMENTS 2A–2C
The motivationfor Experiments2A–2C was to address tual weight of each object in the mental simulation exer-
the possibilitythat demand characteristicsplayed a role in cise played an important role in determininghow quickly
producingthe effect reported in Experiment 1 and to ma- people traveled mentally. To investigate how the amount
nipulatevariables that might affect the amount of weight of weight people imagined lifting would affect their esti-
that people would claim to be able to lift. We attempted mates, some of our participantsimaginedlifting20 pounds,
to reduce demand characteristics by leading the partici- whereas other participants imagined lifting a more sub-
pants to believe that they would take part in two separate stantial amountof weight (200 pounds)or no weight at all
research projects, whereas in truth, the two projects were (a control condition).A secondary motivationfor this ex-
related. The peoplefirst simulatedliftingweights, and then periment was to further investigatethe demand character-
they reported aspects of their imagery experience. The istics account. It is possible that performing any mental
second study was a health survey. This survey contained simulation,regardless of the details, causes an inflationof
one critical item that asked the participantsto estimate how performance estimates. If this is true, people who engage
much weight they thoughtthey could lift. If demand char- in simulationsthat differ in magnitudeshouldprovidesim-
acteristicsare thesole explanationof the inflatedestimates ilar weightestimates. If, however, the amountof weight that
in Experiment 1, disguising the relationship between the peoplesimulatehas an influenceon their weight estimates,

MENTAL SIMULATION
375
people who imagine lifting a great deal of weight should the weights? Was there someone spotting you? Was there a sweaty
provide higher weight estimates than do people who sim- smell in the gym? Was it warm in the gym? Were you breathing heav-
ily after returning the bar to the bench? In addition to these questions,
ulate lifting less weight.
the participants also rated the vividness of their mental simulation
We conductedExperiment2C to determinethe extentto
experience on a scale ranging from 1 to 10 (1, extremely poor; 10,
extremely good).
The participants in the 5- and 10-simulation conditions then com-
which people use the amount of simulated weight as a ref-
erence point for their estimates. By this account, a mis-
2
attributionof familiarity does not influence performance pleted a 16-item survey that asked them to provide some basic demo-
estimates. Rather, peoplesimply use the amountof weight graphic information (e.g., age, gender), as well as some health re-
lated information (Do you smoke? Do you exercise on a regular basis?
in the simulation exercise to anchor their estimates. Par-
Compared with the average student, do you think that you are less
ticipants who simulate lifting 200 pounds may start at
healthy or more healthy? How many times have you visited the doc-
2
00 pounds and then shift their estimates downward to a
tor this year?). The critical survey item, which was the seventh ques-
tion on the survey, asked the participants to estimate how much weight
more manageable weight. Likewise, participants who
simulate 20 pounds may shift upward to a more realistic they thought they could lift. People were allowed 5 min to complete
weight. If this account is accurate, people who simulate a each survey. Following completion of the health survey, the exper-
imenter asked the participants in the 5- and 10-simulation condi-
larger amount of weight should always report that they can
tions whether they noticed the relationship between the two studies
lift more weight. To evaluatethis anchoringhypothesis,we
and to describe the exact nature of this relationship. The people in the
compared the weight estimates from a group who imag-
0
-simulation condition did not experience any simulation, and there-
ined lifting200 poundswith thosefrom a group who imag-
ined lifting 400 pounds. If people are anchoring their es-
fore these participants only completed the health survey.
The participants in Experiment 2B listened to the mental simula-
timates, people in the 400-pounds condition should start tion exercise two times, completed the imagery survey, and then
with a higher value before adjusting to a more realistic completed the health survey. The participants in the 20-pounds con-
dition mentally simulated lifting 20 pounds, whereas the participants
weight estimate. Because they start with a higher value, it
in the 200-pounds condition simulated lifting 200 pounds. Other than
is likely that the adjustment process will produce a higher
the change in weight, these two conditions were identical. The par-
ticipants in the control condition did not perform a mental simula-
weight estimate. Therefore, if anchoringis a viable expla-
nation, people in the 400-pounds condition should pro-
vide a higher weight estimate than do people in the 200-
pounds condition.
tion exercise and, therefore, only filled out the health survey.
In Experiment 2C, we used the same procedures and materials as
those used in Experiment 2B, except that the participants in the 200-
pounds condition mentally simulated lifting 200 pounds, whereas the
participants in the 400-pounds condition simulated lifting 400 pounds.
As with the other control conditions, these participants did not
imagine lifting any weight.
Method
Participants. In Experiment 2A, 82 undergraduates were ran-
domly assigned to perform 0 simulations (n 5 27), 5 simulations
(
n 5 22), or 10 simulations (n 5 33). In Experiment 2B, 129 under-
graduates were assigned to the control (n 5 44), 20-pounds (n 5
Results and Discussion
3
4
5), or 200-pounds (n 5 40) condition. In Experiment 2C, 119 un-
dergraduates were randomly assigned to the 200-pound (n 5 36),
00-pounds (n 5 36), or control (n 5 47) condition. In all three ex-
periments, the participants were tested in small groups of 8–10, and between the imagery and the health survey studies. It is
Experiment 2A. None of the participantsin the 5- and
1
0-simulation conditionsreported detectinga relationship
4
none of them had taken part in Experiment 1.
therefore unlikely that these people intentionally inflated
their weight estimates on the health survey. The partici-
pants in the 0-simulations condition estimated that they
could lift 88 pounds (SE 5 9.09), the participantsin the 5-
simulationsconditionestimated111 pounds(SE 5 17.16),
and the participants in the 10-simulation condition aver-
Materials and Procedure. At the outset of the experimental ses-
sion, the participants in the simulation conditions were informed that
they would take part in two separate studies. The first study was an
investigation of how people mentally simulate an event, and the sec-
ond study consisted of a health survey. In order to increase the ef-
fectiveness of the cover story, one experimenter conducted the men-
tal simulation study, and a second experimenter administered the aged 132 pounds(SE 5 13.4). The overallanalysisof vari-
health survey.
In Experiment 2A, the participants in the 5- and 10-simulation
conditions listened to the following minute-long audiotaped in-
structions:
ance (ANOVA) fell just shy of conventionalstatisticalsig-
nificance [F(2,79) 5 2.8, p 5 .06]. However, a significant
linear trend was present [F(1,79) 5 5.7, p , .02]. Those
participantswho completed more mental simulation trials
believed that they could lift more weight than did the par-
ticipants who completed fewer simulations. These results
fit nicely with the misattribution account. In this case, in-
creasing the number of simulation trials causes the details
of the simulation to come to mind more easily, which in
turn causes people to inflate their weight estimates. These
results are consistentwith Goff and Roediger (1998), who
reported that the number of memory distortionsincreased
as the number of imaginingsincreased.
Close your eyes, feel confident that you will be able to lift the 200 pound
barbells, mentally picture yourself lying on the gray weight bench with
the bar above your head, concentrate on lifting the bar while blocking out
other sights and sounds, take a look at the two 100-pound weight plates,
grab the bar, take a deep breath and gently lift it off the stand, gradually
thrust it in the air and breathe out, return the bar to the stand and get up
from the bench.
The mental simulation instructions were repeated 5 or 10 times,
with a short rest interval between each repetition. The participants then
completed the 12-item survey that assessed the qualitative aspects
of their simulation experience and further supported the notion that
this was an investigation of their mental simulation experience. For
At least two additionalaspects of these data are partic-
instance, people were asked the following questions. What color were ularly noteworthy. First, if demand characteristics were a

3
76
LANDAU, LIBKUMAN, AND WILDMAN
viable explanationof the estimationinflationfound in Ex- parisons indicated that the 200- and the 400-pounds
periment 1, people in the 10-simulation condition would groups were similar (t , 1) but that they both differed
have to be aware that another condition existed, and then from the controlgroup (both ts . 2.36). If the participants
they would have to intentionally increase their estimates. were blindly using the amount of weight as a reference
However, none of the participants in either of the simula- point with which to anchor their predictions, the partici-
tion conditions noticed the relationship between the two pants in the 400-poundsconditionshould have provided a
studies, and because we used a between-subjectsdesign, it higherweight estimate. The fact that the peopledid not do
is unlikelythat they knew that other people were perform- this shows that they assess the plausibility of the contents
ing the simulationonly five times. Therefore, it is unlikely of the mental simulation and that this plays some role in
that the peoplein the10-simulationsconditionconsciously how people use that information to make their estimates.
inflated their estimates. In our view, the participants be-
lieved that the weight estimates reflected the amount of
weight they were capable of lifting. Second, the partici-
pants in the 0-simulation condition provided the same
EXPERIMENT 3
In Experiment 3, we addressed the generalityof the ef-
weight estimate as did the participantsin the control group fect reported in the previous four experimentsby using an
of Experiment 1 (88 pounds). Because neither group en- experimental design modeled after Garry et al.’s (1996)
gaged in mental simulation, these data provide convergent imaginationinflationparadigm. The purpose of this exper-
evidence for baseline performance in this population.
iment was twofold. First, we wanted to determinewhether
Experiment2B. As in Experiment2A, noneof the par- people would alter their performance estimates after en-
ticipantsrecognizedthe relationshipbetween the two stud- gaging in a mental simulationexercise. We used a within-
ies. The weight that theparticipantsmentallysimulateddid subjectsdesignthat allowed us to measure people ’s beliefs
influence their weight estimates [F(2,126) 5 5.63, p , abouttheirabilitiesbefore and after the mental simulation
.
05]. Those participants who did not perform the mental exercises. Second, we examinedwhethermentalsimulation
simulation predicted that they could lift 83 pounds (SE 5 would have the same enhancing effects for behaviors other
.25), whereas the participantsin the 20-poundscondition than the weightlifting estimation used in the four previ-
7
predicted 61 pounds (SE 5 7.28) and the participants in ous experiments. We had the participants complete a 10-
the 200-pounds condition predicted 97 pounds (SE 5 item questionnairethat assessed their abilities for several
8
.42). Planned comparisons revealed that the participants sports-related activities(e.g., shootingfree throws, throw-
in the 20-pounds condition reported they could lift less ing darts, etc.). The participants returned after a 1-week
weight than did the participants in the control condition delay and mentally simulated half of the activitiesbefore
[
t(87) 5 2.14, p , .05] and the participants in the 200- completing the questionnaire for a second time. We pre-
pounds condition [t(83) 5 3.24, p , .05]. Although the dicted that the performance estimates for the simulated
participants in the 200-pounds condition estimated that activities would increase and those for the nonsimulated
they couldlift more weight than the participantsin the con- activities would remain stable. A significant interaction
trol condition (14 pounds), this result was not significant between thetest (pretest or posttest)and the item type (sim-
[
t(82) 5 1.25, p . .05]. Consistentwith the misattribution ulated or not simulated) was expected.
account, the detail of the mental simulation exercise has
predictable effects over the subsequent weight estimates.
Because most people do not know exactly how much
Method
Participants, Materials, and Procedure. Twenty-three under-
weight they can lift, the content of the simulation exercise graduates received extra credit in exchange for their participation. As
in the previous experiments, we tested all the participants in small
serves as a source of information for people to use when
groups, and none had participated in the four previous experiments.
making their weight estimates. Therefore, it is not surpris-
During the first session, we informed the participants that we were
interested in their abilities at certain athletic activities. We asked
them to complete a 10-item questionnaire (see the top half of Table 1
ing to find that the mental simulation exercise can either
increase (Experiments 1 and 2A) or decrease performance
estimates (Experiment 2B). The results of this experiment for selected questions). They were encouraged to answer each ques-
indicate that when the participantsmentally simulatedlift- tion to the best of their ability, making the best estimate possible if they
had never performed one of the activities or were otherwise unsure
ing a small amount of weight, they provided lower weight
about their expected performance. None of the participants said that
estimates than did the participantsin a controlconditionor
they were unfamiliar with the activities.
the participants who simulated a large amount of weight.
The participants returned for the second experimental session fol-
lowing a 1-week delay. The 10 questionnaire items were randomly
Experiment 2C. As with the two previous experi-
ments, none of the participantsin the experimentalcondi- divided into two subsets, each containing five activities. The partic-
tions recognized the relationshipbetween the imagery ex- ipants were randomly assigned to one of the two subsets. They were
asked to mentally simulate the five critical activities in that subset,
ercise and the health survey. The participants in the
whereas the other activities served as a control. For example, one group
control condition estimated that they could lift 82 pounds
of participants simulated the five activities in Subset A (walking,
(
1
SE 5 8.56), whereas the people in the 200-pounds(M 5
14, SE 5 8.9) and 400-pounds (M 5 117, SE 5 11.4)
putting [golf], throwing a baseball [speed], soccer goalie [number of
blocks], and bench pressing weights) and did not simulate the five
conditions estimated that they could lift substantially activities in Subset B (shooting free throws, throwing a football [dis-
more weight [F(2,116) 5 4.32, p , .05]. Planned com- tance], kicking an extra point, throwing darts [number of bull’s-eyes],

MENTAL SIMULATION
377
Table 1
analyzedwith a 2 (test, pretest vs. posttest)3 2 (item type,
simulated vs. not simulated)repeated measures ANOVA.
Overall, the performance estimates increased over the 1-
week delay [F(1,22)5 11.28,p , .05]. The item type main
effect did notreach statisticalsignificance[F(1,22)5 1.55,
p 5 .22]. The test 3 item type interactionwas significant
Selected Questions From the Sports Questionnaire
and a Sample Imagery Exercise Used in Experiment 3
If you were to shoot 10 free throws on a regulation basketball court,
how many would go in?
0
–1–2–3–4–5–6–7–8–9–10 baskets
How far can you throw a football?
–5–10–15–20–25–30–35–40–45–50–55–60–65–70–75+ yards
How high is your vertical leap?
[
F(1,22) 5 4.73, p , .05]. Post hoc analysesindicatedthat
0
the performance estimates for the simulated activities
increased [t(22) 5 3.82, p ,.05], whereas the estimates
for the not-simulated activities did not increase [t(22) 5
1.54, p . .05].
0
–6–12–18–24–30–36–42–48–54–60–66–72 inches
Putting
close your eyes
It is possiblethat we failedto detecta significantincrease
for the not-simulatedevents because of a Type II error. To
address this concern, we performed a power analysis on
both post hoc comparisons. The power values for the sim-
ulated and not-simulated activities were .954 and .314,
respectively, and the effect sizes were .399 and .097. Al-
though it is possible that the not-simulated activities in-
creased during the intervening week and that we are not
detectingthis effect because of a power problem, the data
suggest that this increase is quite small. Thus, we would
need hundreds of participants to increase power in order
to detectsuch a small effect. We believethat the interaction
term demonstratesthat the simulated activitiesincreased
more than did the not-simulated activities. Simulatingan
event has some effect on people ’s estimates of their future
performance. The other point that we wish to make is that
feel confident that you will be able to make the putt
mentally picture yourself with both feet flat on the green grass
see the hole as being very large
concentrate on swinging the club straight back and hitting the ball
solidly without other sights and sounds
imagine that the ball travels directly to the hole and plops into the cup
feel very confident and happy that the ball went into the cup
What color was your club?
Was there a flagpole sticking out of the hole?
Were you nervous that the ball might miss the cup?
and vertical leap). A second group of participants simulated the five
activities in Subset B but did not simulate the five activities in Sub-
set A.
The participants were informed that they would complete the five
mental simulation exercises that were described in their experimental
packets. Each description contained a series of steps for the simulation our results are consistentwith the existing imaginationin-
exercise (see the bottom half of Table 1 for an example). Below each flationliteraturein that the imaginedeventsshoweda larger
described activity, there were three questions about what they had sim-
increase than the not-imagined items (e.g., Garry, Shar-
ulated. These questions were designed to ensure that the participants
man, Wade, Hunt, & Smith, 2001; Heaps & Nash, 1999;
complied with the task instructions.
In an effort to give people practice with mentally simulating an
event, instructions for a sample mental simulation exercise were read
aloud:
Pezdek & Eddy, 2001). In the imagination inflation litera-
ture, the not-imagined items tended to increase, but this
increase was typically smaller than the increase associ-
ated with the imagined items.
Close your eyes; mentally picture yourself standing in the doorway of your
home. As you enter the house take a look around and begin counting the
windows. Slowly make your way to the different rooms and count the win-
dows in each room. After visiting all of the rooms, return to the doorway
and exit the house.
The results from this experiment indicate that mentally
simulating an activity increased performance estimates.
Thisfindingis consistentwith the resultsfrom the between-
subjects designs (i.e., Experiments 1–2C). The results
from Experiment 3 are also meaningful,because they help
to rule out demand characteristics as a viable explanation
for these effects. In order for demand characteristics to
fully explain these results, our participants would have
Following the practice exercise, the participants wrote their answers
to the following questions regarding their experience. How many win-
dows do you count? Was it sunny outside? What was the temperature
inside the house?
The participants had 60 sec to complete each simulation exercise.
The participants then had 30 sec to write answers to the three ques- needed to remember their exact ratings for all 10 activi-
tions. This procedure was repeated for the five target activities. After ties from the first week. Then, during the second week,
all five exercises were completed, the experimenter announced that
they would have to boost their ratings for the simulatedac-
the sports activities questionnaires that they completed during the first
tivitieswhilekeepingthe ratingsfor the otheractivitiescon-
week had been accidentally misplaced, and the participants were then
sistent. Because it is unlikelythat people remembered the
asked to complete the questionnaire a second time. Following admin-
istration of this posttest, we debriefed and dismissed all of the partic-
ipants from the experimental session.
Table 2
Mean Performance Ratings (With Standard Error)
Results and Discussion
We convertedparticipantresponsesto proportionsbased
on the highest level of performance for the simulated and
not-simulated questionnaireitems (see Table 2). For exam-
ple, for the free throw question, we dividedeach response
by 10, and for the verticalleapquestion,we dividedeach re-
sponse by 72. These proportions were then averaged and
for Experiment 3
Session
Week One
Week Two
Item Type
M
SE
M
SE
Not-simulated
Simulated
.31
.33
.02
.03
.34
.39
.02
.03

3
78
LANDAU, LIBKUMAN, AND WILDMAN
exact rating for each of the activities, we doubt that de- offered higher (Experiment2C) or lower (Experiment2B)
mand characteristicsprovidea completeexplanationof our estimates for the imagined tasks. The results from Exper-
results. If demand characteristics had inflated the weight iment 2C are interestingbecause they show that peopledo
estimates, only the main effect for session, and not the not blindly anchor their weight estimates to the values
two-way interaction, would have been significant.
mentioned in the mental simulation exercise. Like other
misattribution-of-familiarityeffects, peopleappearto have
some cognitivecontroloverthis error (e.g., Whittlesea,Ja-
coby, & Girard, 1990). A similar misattribution of avail-
GENERAL DISCUSSION
The purpose of this research was to determine whether a ability might cause the inaccuracy of the immediate JOLs
brief mental simulationexercise influencesprospectivees- (cf. Nelson & Dunlosky,1991). Specifically, when people
timates for physicalabilities.After peopleimagined them- attempt to immediately predict future memory perfor-
selves lifting a heavy object, they claimed that they could mance for a target word, they tend to rely on the availabil-
lift more weight than did people who had not engaged in ity of the word to make their judgment.Because the study
a mental simulation exercise (Experiment 1). Increasing pair still resides in short-term memory, this interferes
the frequency of the simulationexercise resulted in higher with people ’s ability to accurately predict future recall of
estimates (Experiment 2A). People who simulated lift- the target word. The common thread in these two cases is
ing a larger amountof weight estimatedthat they couldlift that informationthat easily comes to mind influencedsub-
more weight than did people who simulated lifting a sequent judgments.
smaller amount of weight (Experiment 2B). Importantly,
In terms of future research, we suspect that several fac-
we demonstrated that anchoring was not a viable expla- tors will influence the magnitude of the estimation infla-
nation of estimation inflation, because people who men- tion. For example, if the simulation exercise is concrete
tallysimulated200 or 400 poundsprovided similar weight and incorporates tangible details, the weight estimates
estimates (Experiment 2C). Finally, people increased should increase. People who imaginerealistic and/or plau-
their performance estimates for activities that they men- sible events should offer performance estimates that out-
tally simulated, whereas activities that people did not weigh those made by people who imagine less realistic
simulate remained relatively stable (Experiment 3).
Several aspects of the results from these experiments
events (e.g., Pezdek, Finger, & Hodge, 1997).
Mental simulationis a powerful techniquethat benefits
raise serious doubtsthat the inflatedweight estimates were mnemonic (Yesavage, Rose, & Bower, 1983) and athletic
a result of demand characteristics.First, Experiments2A– (Wrisberg & Anshel, 1989) performance. However, re-
2
C used a procedure that disguised the relationship be- search indicatesthat mental simulationexercises can yield
tween the mental simulation exercise and the weight es- unintendedside effects. Mentallysimulatingthat an event
timate, and none of the participants in these experiments has taken place, when it has not, causes people to believe
reported that they were aware of the relationshipbetween that the event actually occurred (Garry et al., 1996). Sim-
the two studies. Second, if demand characteristicswere an ulations also have the ability to influence attitudes and
accurate explanation of these results, the participants in behavior(Gregory et al., 1982). Our results are important
the different experimentalconditionsshouldhave provided because they provide evidence that physical ability judg-
roughly the same weight estimates. For instance, the re- ments are constructed much like our recollections of the
sults from Experiment 2B demonstrate that the partici- past and that they are similarly malleablefollowing a brief
pants in the 200-poundsconditionpredictedthat they could mental simulationexercise. Thus, peoplewho use mental
lift more weight than did people in the 20-pounds condi- simulationexercises to monitor their performance should
tion. It is unclear why the participants in the 200-pounds be careful to make sure that they are not over- or under-
condition would be compelled to assist the experimenter estimating their actual performance level.
more than would peoplein the 20-poundscondition.Third,
mental simulationsaffected weight estimates in between-
subjects designs (Experiments 2A–2C) and a within-
subjects design (Experiment 3).
The results from the present experiments demonstrate
that even a brief and impoverishedmental simulation ex-
ercise is a powerful tool that can influence people ’s ability
to predictfuture physicalperformance. Althoughour para-
digm required prospectiveperformance estimates, rather
than recollectionsof past experiences,we suspect that the
REFERENCES
Dunlosky J., & Nelson, T. O. (1994). Does sensitivity of judgments of
learning (JOLs) to the effects of various study activities depend on
when the JOLs occur? Journal of Memory & Language, 33, 545-565.
Garry, M., Manning, C. G., Loftus, E. F., & Sherman, S. J. (1996).
Imagination inflation:Imagininga childhoodevent inflates confidence
that it occurred. Psychonomic Bulletin & Review, 3, 208-214.
Garry, M., Sharman, S. J., Wade, K. A., Hunt, M. J., & Smith, P. J.
2001). Imaginationinflation is a fact, notan artifact: A reply to Pezdek
and Eddy. Memory & Cognition, 29, 719-729.
(
same cognitive processes produce imagination-induced Goff, L. M., & Roediger,H. L., III (1998). Imagination inflationfor ac-
distortionsfor bothtypesof judgments.In bothcases, peo-
ple’s judgments of an ambiguous situationare influenced
by the most availableinformation.In the present case, peo-
ple failed to consider that their weight estimates were in-
fluencedby the simulationexercise,and consequently, they
tion events: Repeated imaginings lead to illusory recollections. Mem-
ory & Cognition, 26, 20-33.
Gregory, W. L., Cialdini, R. B., & Carpenter, K. M. (1982). Self-
relevant scenarios as mediators of likelihood estimates and compli-
ance: Does imaginingmake it so? Journalof Personality & SocialPsy-
chology, 43, 89-99.

MENTAL SIMULATION
379
Heaps, C., & Nash, M. (1999). Individualdifferences in imagination in- Pezdek,K., & Eddy, R. M. (2001).Imaginationinflation:A statistical ar-
flation. Psychonomic Bulletin & Review, 6, 313-318.
Hyman, I. E., Jr., & Billings, F. J. (1998). Individualdifferences and the
creation of false childhood memories. Memory, 6, 1-20.
tifact of regression toward the mean. Memory & Cognition, 29, 707-717.
Pezdek, K., Finger, K., & Hodge, D. (1997). Planting false childhood
memories: The role of event plausibility. Psychological Science, 8,
437-441.
Intons-Peterson,M. J., & Roskos-Ewoldsen, B. B. (1989). Sensory-
perceptual qualities of images. Journal of Experimental Psychology: Pham, L. B., & Taylor, S. E. (1999). From thought to action: Effects of
Learning, Memory, & Cognition, 15, 188-199.
Jacoby,L. L., & Kelley,C. M. (1987).Unconsciousinfluences of mem-
process- versus outcome-based mental simulations on performance.
Personality & Social Psychology Bulletin, 25, 250-260.
ory for a prior event. Personality & Social Psychology Bulletin, 13,
Ross, M., & Sicoly,F. (1979). Egocentric biases in availability and attri-
bution. Journal of Personality & Social Psychology, 37, 322-336.
3
14-336.
Johnson, M. K. (1988). Reality monitoring: An experimental phenom- Taylor, S. E., Pham, L. B., Rivkin, I. D., & Armor,D. A. (1998). Har-
enological approach. Journal of Experimental Psychology: General,
17, 390-394.
nessing the imagination: Mental simulation, self-regulation, and cop-
ing. American Psychologist, 53, 429-439.
1
Johnson, M. K., Hashtroudi, S., & Lindsay, D. S. (1993). Sourcemon- Whittlesea, B. W. A., Jacoby, L. L., & Girard, K. (1990). Illusions of
itoring. Psychological Bulletin, 114, 3-28.
immediate memory: Evidence of an attributional basis for feelings of
familiarity and perceptual quality.Journalof Memory & Language, 29,
716-732.
Wrisberg,C. A., & Anshel, M. H. (1989).Theeffects of cognitivestrate-
gies on the free throw shooting performance of young athletes. Sport
Psychologist, 3, 95-104.
Yesavage,J. A., Rose,T. L.,& Bower, G. H. (1983).Interactive imagery
and affective judgments improve face-name learning in the elderly.
Journal of Gerontology, 38, 197-203.
Johnson. M. K., Raye,C. L., Hasher, L., & Chromiak, W. (1979). Are
there developmental differences in reality-monitoring? Journal of
Experimental Child Psychology, 27, 120-128.
Kelley,C. M., & Jacoby,L. L. (1996). Adult egocentrism: Subjectiveex-
perience versusanalytic bases forjudgment.Journalof Memory & Lan-
guage, 35, 157-175.
Kelley, C. M., & Lindsay, D. S. (1993). Remembering mistaken for
knowing:Ease of retrieval as a basis for confidence in answers to gen-
eral knowledge questions. Journal of Memory & Language, 32, 1-24.
Landau, J. D., & Marsh, R. L. (1997). Monitoringsource in an uncon-
scious plagiarism paradigm. Psychonomic Bulletin & Review, 4, 265-
NOTES
2
70.
1
. We removed one statistical outlierfrom the data analysis, leaving 43
Marsh, R. L., & Landau, J. D. (1995) Item availability in cryptomnesia:
Assessing its role in two paradigms of unconsciousplagiarism. Jour-
nalof Experimental Psychology:Learning,Memory, & Cognition, 21,
participants in both conditions.Because our hypothesiswas directional,
we report one-tailed significance tests in this experiment.
2
explanation.
3
experiment after we collected the data from the 20-poundand 200-pound
groups. To ensure some type of random assignment, we also added par-
ticipants to the 20- and 200-poundconditions.It is important to note that
the additional data did not substantially alter the means or standard er-
rors for these two groups.
. We would like to thank Deanne Westerman for raising this potential
1
568-1582.
Marsh, R. L., Landau,J. D., & Hicks, J. L. (1997). The contributionof
inadequate source monitoringto unconsciousplagiarism during idea
generation. Journal of Experimental Psychology: Learning, Memory,
. Followinga reviewer’s suggestion, we added a control group to this
&
Cognition, 23, 886-897.
Nelson,T. O.,& Dunlosky, J. (1991).When people ’s judgmentsof learn-
ing (JOLs) are extremely accurate at predicting subsequentrecall: The
“
delayed-JOL effect.” Psychological Science, 2, 267-270.
Paddock,J. R.,Joseph,A. L.,Chan,F. M.,Terranova,S., Manning,C.,
Loftus, E. F. (1998). When guided visualization procedures may
&
backfire: Imagination inflation and predicting individualdifferences
(Manuscript received March 18, 1999;
in suggestibility. Applied Cognitive Psychology, 12, 63-75.
revision accepted for publication January 9, 2002.)