Memory markers: How consumers recall the duration of experiences
Hee-Kyung Ahn, Maggie Wenjing Liu, Dilip Soman⁎
Rotman School of Management, University of Toronto, Canada
Received 27 August 2007; accepted 27 August 2007
Available online 12 June 2009
In the present article, we propose a three-stage memory marker model of memory for experience. The human mind generates and encodes
“memory markers”of specific episodes, stores them in memory, and after a temporal delay retrieves these markers to reconstruct the experience
and make relevant judgments. Rich experiences characterized by vivid stimuli seem to pass by quickly, yet feel longer when recalled after a period
of time because the number of retrieved memory markers is large. We also examine situations in which key predictions of the memory marker
model can be moderated. A field study and five laboratory experiments were conducted to test various aspects of the memory marker model and
provide process support.
© 2009 Society for Consumer Psychology. Published by Elsevier Inc. All rights reserved.
The field of marketing has evolved away from the discipline
of selling products to the discipline of creating experiences for
consumers (Schmitt, 2003). In this article, we focus on one
important aspect of experiences that influences consumer
satisfaction and subsequent decision-making —its duration
(e.g., Antonides, Verhoef, and Aalst, 2002).
Past research suggests that subjective duration judgments of
objectively identically long experiences might be influenced by
changes in the environment (e.g., Kellaris and Kent, 1992). In
particular, periods of time spent in rich environments with lots
of distractions might be seen to have passed by quickly. The
subjective duration of an experience has been measured either
as perceived speed (e.g., “How quickly do you think time was
passing by”) or as perceived duration (e.g., “what is your best
estimate of the duration?”). The simplest and most intuitive
relationship between these two variables is that the faster time
seems to pass, the smaller should the duration of the experience
be. We propose, however, that rich experiences in which time
seems to move quickly might also be reported to be longer in
duration when recalled after a temporal delay. If time passes
quickly in a rich experience, why should a consumer believe it
had a greater duration when asked after a delay?
We resolve this seeming discrepancy by proposing that
duration judgments made after a delay arise from a process that
is fundamentally different from judgments made online or
immediately at the conclusion of the episode. In particular, we
propose a model of consumers' memory for experiences that we
call the “memory marker model.”Similar to the Czech author
Milan Kundera who wrote that “memory does not make films, it
makes photographs”(Kundera, 1999) we propose that the
human brain generates mental memory markers of the
environment when there are cognitive or sensory changes that
occur around us. These markers could be visual (for instance,
the memory of a beautiful painting or a human face), auditory
(the memory of the notes of one's favourite symphony piece),
tactile —and indeed occur in any sensory or cognitive form.
These memory markers are categorized and stacked in memory
bins, much like sheets of paper in an office tray (Wyer and Srull,
vailable online at www.sciencedirect.com
Journal of Consumer Psychology 19 (2009) 508 –516
E-mail addresses: firstname.lastname@example.org (H.-K. Ahn),
email@example.com (M.W. Liu),
Dilip.Soman@rotman.utoronto.ca (D. Soman).
Hee-Kyung Ahn and Maggie Wenjing Liu are Ph.D. students in marketing
and Dilip Soman is the Corus Chair in communications strategy and professor of
marketing at the Rotman School of Management, University of Toronto, 105 St.
George Street, Toronto, ON M5S 3E6, Canada. This research is supported by a
grant from the Social Sciences and Humanities Research Council of Canada
(SSHRC) and the Desautels Center for Integrative Thinking (DCIT) at the
Rotman School of Management to the third author. All authors contributed
equally to the research reported in this paper, and the order of authorship is
alphabetical. We thank Shane Frederick, Hae-Joo Kim, Andy Mitchell, Darlene
Walsh, Bob Wyer, Gal Zauberman, Meng Zhang, and seminar participants at
HKUST, McGill University, Concordia University, and the University of British
Columbia for comments, suggestions and insightful discussion. We thank Edbert
Khong and Edrea Khong for inspiring the study reported in the conclusion
section. We also thank Rohit Balakrishna, Patrice Blanc, Emily Chow, Ailing
Chua, Sandesh Kulkarni, Mandy Li, Szeling Tam and Sumit Vyas for excellent
research assistance. Correspondence should be addressed to Dilip Soman.
1057-7408/$ - see front matter © 2009 Society for Consumer Psychology. Published by Elsevier Inc. All rights reserved.
1989). While recalling events, consumers retrieve the memory
markers from the appropriate bin and view them to retrieve the
experience. The number of memory markers serves as a cue to
delayed duration judgment. Consequently, in the case of rich
stimuli where the environment changes frequently (like the
themed ride), there are significantly more memory markers,
resulting in a longer duration judgment.
We further show that factors that inhibit encoding or re-
trieving (e. g., ego-depletion and cognitive load) might limit the
number of markers encoded and retrieved respectively, and
hence shorten the delayed duration judgment. On the other hand,
factors that facilitate the encoding of markers (e.g., sensitivity to
changes in the environment) should increase the number of
markers recalled, hence prolonging the estimated duration.
The rest of this paper is organized into three sections. First,
we review relevant literature and describe our proposed three-
stage memory marker model. We also examine situations in
which the key predictions of the memory marker model can be
moderated. Second, we describe a field study and five
laboratory experiments designed to test various aspects of the
memory marker model and provide process support. Finally, we
conclude with a general discussion and propose directions for
The memory marker model: a theoretical framework
Research has studied various aspects of the memory of past
experiences. One such aspect relates to the gestalt evaluation of
experiences at their conclusion. Fredrickson and Kahneman
(1993) found that salient features of affective experience (e.g.,
the best or worst part of the experience, the rate of change of the
experience, and the end state) seem to influence judgment more
than their objective duration, and called this phenomena
“duration neglect”(see also Kahneman, Fredrickson, Schreiber,
and Redelmeier, 1993; Redelmeier and Kahneman, 1996).
Clearly, right at the end of the experience, the salient features of
the experience are immediately available as an input into
judgment, while an “evaluation that incorporates duration must
be constructed more laboriously”(Ariely, Kahneman, and
Loewenstein, 2000, p. 524).
Another aspect of the memory for experiences relates to
how individuals judge its duration (e.g., Block, 1990; Block
and Zakay, 1997; Glicksohn, 2001; Stroud, 1967; Treisman,
1984). There are three common themes in prior work in the
area of time perception. First, the work explicitly or implicitly
implies the existence of an internal timer to mark the passage
of time. Second, duration judgment is often flexible and
inaccurate (as compared to objective time), and influenced by
multiple factors like emotions, mental engagement, novelty,
and variety in activity, among others. Third, prior models are
mostly paramorphic in nature and do not provide much
evidence for the underlying processes.
In our research, we study the duration judgment of an
experience made after a significant delay (in our experimental
work, the delay is to the order of a few days). Unlike past
research which uses the term “retrospective judgment”to refer to
a judgment made immediately at the conclusion of an
experience, we use “delayed judgment”to refer to a judgment
made after a significant delay such that visceral factors and
salient features of the experience are not immediately accessible.
Before describing our memory marker model, we offer a
metaphor to illustrate its basic features. Imagine that last year
Laurel and Hardy each went on an identical trip with family.
Laurel took photographs of every family member he met and
every event he participated in. Hardy did not take as many
photographs. Several months later, when both men were asked to
recall this otherwise unremarkable trip, they each went back to
their digital photo album marked “family trip”and viewed their
trip pictures as a slideshow. After viewing his 100 photographs,
Laurel seemed to recall that he had a longer and more eventful
trip than did Hardy who had viewed his 20 photographs.
This story captures the basic intuition underlying our memory
marker model. The following elements of the story are key
variables in our model: a) photographs that served as memory
markers were taken by the two men during their trip, b) the
photographs were all stored in an album which was tagged with
the name of the experience, and c) the number of photographs
was used as a cue to judge duration. These correspond to the
three stages of our memory marker model: the encoding of
markers, filing them in appropriate memory bins, and retrieving
them after a delay.
Encoding refers to the process by which certain slices or
moments of the experience are recorded for later retrieval,
typically when the environment changes. Not all changes in our
cognitive and sensory environment result in a marker. For
instance, Laurel and Hardy might watch the same exact movie
and encode different parts of the movie. These differences could
be due to differences in attention caused by preoccupation with
other thoughts or due to competing stimuli (lower attention
would result in a smaller number of memory markers),
differences in whether each stimulus is particularly unusual for
each consumer (unusual stimuli are more likely to be encoded as
memory markers), or differences in the rate of change of the
stimulus (frequent changes are more likely to result in the
encoding of a larger number of memory markers).
In the filing stage, the encoded memory markers are
categorized and stacked in memory bins, a process consistent
with Wyer and Srull's (1986, 1989) bin model of memory. This
model conceptualizes memory as bins, or storehouses of
individual pieces of information about people, objects, and
events. The output of information processing is transmitted to
and stored in a relevant bin as a separate information re-
presentation, in the order it is generated.
During the retrieving stage, the memory markers stored in
memory bins are reviewed and used to make judgments. Given
that a larger number of markers would result in greater time
needed to review them, we expect that the number of memory
markers will serve as a cue to infer the duration of the recalled
experience. Note, however, when an experience is rich in
cognitive and sensory changes, time might have flown by faster
(Kellaris and Kent, 1992). However, such an experience will
result in a greater number of memory markers, and as a result, it
will be remembered as a long experience. This prediction of the
memory marker model is consistent with the paradox noted by
509H.-K. Ahn et al. / Journal of Consumer Psychology 19 (2009) 508–516
William James (1890, Chapter 15) that “in general, a time filled
with varied and interesting experiences seems short in passing,
but long as we look back. On the other hand, a tract of time
empty of experiences seems long in passing, but in retrospect
short.”We use the terms “rich”and “impoverished”to refer to
situations in which a consumer encodes a) a number of memory
markers of “varied and interesting experiences”and b) very few
memory markers from a “tract of time empty of experiences”
respectively and formally hypothesize:
H1. The delayed duration judgment of a rich experience will be
longer than that of an impoverished experience of the same
We also note the conceptual difference between the drivers of
the perceived duration in James' (1890) view and ours. James
notes that the perception of duration is a function of the
properties of the stimulus, while in our model it is the property
of the interaction of the consumer and the stimulus. Specifically,
we believe that a time filled with varied experience may not
appear longer after a delay if the consumer fails to encode the
changes as memory markers.
Factors which influence encoding, filing and retrieving of
markers should also impact our delayed duration judgments.
Factors that inhibit [facilitate] encoding or retrieving might limit
[enhance] the number of markers encoded and retrieved
respectively and hence shorten [lengthen] the delayed duration
judgment. We therefore hypothesize that:
H2. The effect of richness of experience on the delayed
duration judgment of the experience will be weakened when
consumers are under a cognitive load at the time of retrieval.
H3. The effect of richness of experience on the delayed
duration judgment of the experience will be weakened when
consumers are ego-depleted during the experience.
H4. The effect of richness of experience on the delayed
duration judgment of the experience will be enhanced when
consumers are sensitized to changes in the environment during
We next present the results of several studies designed to test
We conducted a field study in two international locations of a
theme park. Participants were compensated with souvenir gifts
and tickets to other local attractions. Participants (and their
families) typically went on a number of rides on a given day,
and had all purchased a three day pass. Data were collected from
each participant at three points in time —a) on the first day of
their visit to the park, we asked them to simply list each of the
rides they went on, b) on the second day, we also asked them to
answer one question at the end of each ride they went on, and c)
at the end of their second day, they were asked questions about
the duration of rides they had gone on for the first day. The
average participant went on 6 rides on each of the two days; and
no participant went on the same ride on the two successive days.
On the second day, at the immediate end of each ride that they
went on, participants were asked to circle a number on a scale to
indicate their response to “How quickly do you think time was
passing by during the ride”(1 = Very slowly, 9 = Very fast). And
at the end of the second day, theywere reminded of the rides they
took on the first day and asked two questions —a) “How long do
you think the ride was?”(HOWLONG: 1 = Very short, 9 = Very
long), and b) “Provide your best estimate of the duration of the
ride”(___ minutes ___ seconds). The dependent measure we
used was the ratio of their estimate to the actual duration
In analyzing our data, we only looked at one factor along
which to compare these different measures; the nature of the
ride. Three of the eight rides we studied were what we
called “repetitive”rides. An example of such a ride is a
carousel, where the motion is primarily circular and the ride
is made up of an aggregation of identical elements (e.g.,
revolutions). The remaining rides were what we call
“storyline”rides. These took the rider from one location to
another, each depicting a new piece of information or part of
a story. In the language of hypothesis 1, the repetitive rides
were relatively impoverished, while the storyline rides were
We first consider the data collected immediately at the end
of each ride. Participants reported that time seemed to have
passed more slowly for the repetitive rides (n= 30, M= 4.57)
than for the storyline rides (n= 29, M= 6.21, t(57) = 2.96,
pb.01). This might have implied that repetitive rides seemed
longer than storyline rides. However, we found the opposite
pattern of data for the rides where evaluations were collected
after a time delay (i.e., for first day rides as recalled at the end
of the second day). For delayed evaluations, participants
reported that repetitive rides appeared shorter (n= 32, HOW-
LONG: M= 4.34; RATIO: M= 1.018) than storyline rides
(n= 31, HOWLONG: M= 5.71, t(61) = 2.45, pb.01; RATIO:
M= 1.306, t(61) = 3.44, pb.01).
These results were consistent with hypothesis 1. That said, we
were also acutely aware of the limitations of field studies in
general. Our goal was to replicate these effects in the laboratory
in a more controlled setting. In all of our experimental studies,
participants viewed stimuli in the form of photographs or video
clips under an appropriate cover story, and were asked to remove
all devices like watches, jewelry, cell phones —absolutely
anything that had any rhythmic component that participants
could use to estimate time duration. This is common in research
on time perception (Block and Zakay, 1997; Vohs and
In this experiment, we created rich vs. impoverished
experiences by exposing our participants to slideshows with
photographs of various landscapes and natural scenery. We used
a particular sequence of photographic slides as the experience
that we wanted participants to recall, and manipulated the
number of slides in the slideshows while holding the actual
510 H.-K. Ahn et al. / Journal of Consumer Psychology 19 (2009) 508–516
Participants, design, stimuli and procedure
Sixty students were paid for participating in this experiment.
They were then told that they would be watching a slideshow
with several photographs in various domains (we used pictures
of fish, flowers, mountains, a lake, a bird, balloons, and a dog).
They were randomly assigned to one of two conditions;
participants in one condition watched a slideshow that consists
of six different slides (i.e., fish, mountain, balloon, bird, horse,
wild flowers) with each slide displayed for 30 s. Participants in
the other condition viewed thirty different slides, with each slide
displayed for 6 s. The slideshow in this second condition
contained all six photographs from the first slideshow, and these
were placed on 1st, 6th, 11th, 16th, 21st, and 26th position
respectively. While the objective duration of both slideshows
was the same, the show with 30 slides was richer since it gave
participants more information to encode.
After watching the slideshow, some participants in the
immediate judgment condition were asked to fill out a
questionnaire that measured their subjective judgment of the
duration of the slideshow. In addition, participants also were
asked to rate their mood, boredom and nervousness
(Depressed–Cheerful, Unhappy–Happy, Bad–Good, Bored–
The remaining participants (in the delayed judgment
condition) were told that they would receive the [same]
questionnaire by e-mail three days later, and were asked to
respond to it upon receipt. This experiment therefore employed
a 2 (slideshow: [6 slides × 30 s] vs. [30 slides × 6 s]) × 2 (timing
of judgment: immediate vs. delayed) between-participant
Results and discussion
To test hypothesis 1, we conducted a 2 (slideshow) × 2
(timing of judgment) ANOVA with duration judgment as the
dependent variable. The results are plotted in Fig. 1, and show a
significant interaction effect of the slideshow with the timing of
evaluation (F(1, 56) = 94.3, pb.001). As the figure shows,
participants in the immediate condition reported a longer
duration when the number of slides was small (M= 208.0)
compared with large (M= 120.7; F(1, 28) = 61.64, pb.001).
However, participants in delayed condition perceived the
duration to be longer when the number of slides was large
(M= 212.7) compared with smal l (M= 151.7; F(1, 28) = 33.94,
pb.001). Mood, nervousness and boredom were used as
covariates in the 2 (slideshow) × 2 (timing of judgment)
ANCOVA and did not influence participants' duration
The results of this experiment were consistent with H1.
When individuals engage in delayed judgments, individuals
who have viewed the larger number slides perceived the
duration of the experience to be longer. This is in contrast to
situations where individuals are asked to make judgments
immediately after the event, where they seem to be drawn by the
richness of the experience and perceive time passing quickly
and duration is short.
In this experiment, we used the number of slides as a
manipulation to induce cognitive changes. In doing so,
participants in one condition simply saw a larger variety of
pictures as compared to participants in the other condition, and
it may be argued that this extra sensory input might influence
duration judgments. A cleaner test would hold the total content
of the pictures constant, and Experiment 2 was conducted with
this in mind.
Participants, design, stimuli and procedure
Sixty students were paid for participating in this experiment.
They saw a total of five separate photographic slides
(specifically, five different species of butterflies varying in
colors, size, etc.). Each repeated five times in either a cyclical or
repetitive manner, with each exposure lasting for 5 s. Hence the
total slideshow lasted 125 s.
For some participants (in the repetitive transition condition),
each slide was displayed for 5 s, and was repeated five times.
We used different animation effects during each successive
transition (e.g., picture flying in from the left, right, bottom, top,
and top left). At the end of the first slide, the next slide was
presented for five repetitions in the same manner till all five
pictures had been shown (specifically, if the slides were labeled
1, 2, 3, 4 and 5, participants saw 11111/22222/33333/44444/
55555). The remaining participants (in the cyclical transition
condition) viewed the slideshow in a sequence that can be
represented as 12345/32154/13524/31542/54321. We used the
same animation effects during each successive slide in the
cyclical transition condition. The cyclical transition condition
was richer because each transition represented a different
After watching the slideshow, participants filled out a
questionnaire either immediately or three days later as the
procedures in experiment 1. This experiment therefore used a
2 (slide transition: repetitive vs. cyclical transition) ×2 (timing
of judgment: immediate vs. delayed) between-participants
Fig. 1. Duration as a function of timing of judgment and slideshow: Experiment 1.
511H.-K. Ahn et al. / Journal of Consumer Psychology 19 (2009) 508–516
Results and discussion
The mean subjective duration judgments are plotted in Fig.
2. A 2 (slide transition) × 2 (timing of judgment) ANOVA
showed a significant two-way interaction between sequence and
timing of judgment (F(1, 56) = 6.6, pb.05). Participants in the
immediate condition estimated a longer duration in the
repetitive transition (M= 123.7) compared to the cyclical
transition (M= 106.7; (F(1, 28) = 3.42, pb.1). In contrast,
participants in delayed condition reported a longer duration
when the slide transition was cyclical (M= 120.7) vs. repetitive
(M= 109.3; F(1, 28) = 3.46, pb.1).
Across the first two experiments, we manipulated the relative
richness of the slideshow experience in two different ways. In
both cases (and in the field study reported earlier), the results
supported H1 and William James' observation (1890, p. 624). In
the following three experiments we tested the moderators on the
delayed duration judgments with regard to the encoding and the
retrieval stage in the memory marker model.
The goal of experiment 3 was to test the memory marker
model at the retrieval stage, and in particular to test H2. Each
participant saw both repetitive and cyclical slideshows at the
first session (rather than this being a between-subjects variable).
Also, a number-memorizing task was used to manipulate the
cognitive load at the time of estimating duration.
Participants, design, stimuli and procedure
In the first session, fifty-eight participants watched two
slideshows (repetitive and cyclical) separated by a filler task.
One slideshow had pictures of birds, while the other had
pictures of butterflies (same stimuli used in experiment 2). The
type of the slideshow (i.e., butterflies vs. birds) and the order of
slideshows (repetitive one first vs. cyclical one first) were
counterbalanced across slide transition. Neither the type of
slideshow nor the order had any effects on the dependent
measures, and hence are not discussed further.
In the second session, participants were randomly assigned
to one of two conditions (high cognitive load vs. no-load).
Participants in high cognitive load condition were asked to
memorize a random 9-digit number (i.e., 937852965) and were
told they if they could recall the number accurately at the end of
the session, they would receive an extra $2. Participants in the
no-load condition were simply asked to fill up the questionnaire.
This experiment therefore employed a single factor —cognitive
load (high vs. no) design.
Participants were asked to indicate which slideshow they
recalled as being longer (LONGER: 1 = Bird, 6= Equally Long,
11= Butterfly). We also measured each participant's best judgment
of the duration on each slideshow. We recoded the data as
the LONGER variable (i.e., 1 = Repetitive, 6 = Not Sure,
11 = Cyclical).
Results and discussion
In order to test hypothesis 2, we conducted a one-way ANOVA
with participants' reported judgment about which slideshow was
longer as the dependent variable, and cognitive load as the
independent variable. The mean of LONGER score indicating
that participants found the cyclical slideshow longer than the
repetitive one in the high cognitive load condition (M=5.5) was
not significantly different from the midpoint of the scale
(“Equally Long”=6, p= .21). In contrast, participants in the no-
load condition indicated that the cyclical slideshow was longer
than the repetitive slideshow (M=7.9; F(1, 56) = 20. 49, pb.001).
As a secondary analysis, we also conducted a 2 (slide
transition: repetitive vs. cyclical transition) × 2 (cognitive load:
high vs. control) ANOVA with the duration judgment of each
slideshow as the dependent variable. Like previous experi-
ments, participants estimated a shorter duration in the repetitive
transition (M= 93.5) of the slideshow than in the cyclical
transition (M= 108.7; F(1, 56) = 12.53, pb.005). Moreover,
those under high cognitive load condition (M= 90.7) estimat ed a
shorter duration of each slideshow than those in the no-load
condition (M= 111.5; F(1, 56) = 7.06, pb.05). The ANOVA
results also showed a significant two-way interaction between
Fig. 2. Duration as a function of timing of judgment and slide transition:
Fig. 3. Duration as a function of cognitive load and slide transition: Experiment 3.
512 H.-K. Ahn et al. / Journal of Consumer Psychology 19 (2009) 508–516
slide transition and cognitive load (F(1, 56) = 9.12, pb.005).
The results are plotted in Fig. 3 and show that participants in the
no-load condition felt the duration of the slideshow with the
repetitive transition (M= 97.4) was shorter than the cyclical
transition (M= 125.5). However, participants in the high
cognitive load condition reported that the repetitive transition
(M= 89.6) and the cyclical transition (M= 91.9) were not
significantly different, implying that the slide transition had
no effect. This pattern of data supported H2. The results in this
experiment show that inhibiting mental capability at the
retrieval stage inhibits the retrieval of memory markers, and
therefore leads to shorter duration judgment.
Participants, stimuli, procedure and measurement
Sixty-four female participants (age 18–60) participated in a
food tasting study in exchange for cash compensation. The two
slideshows used in experiment 2 were fully crossed by a second
factor, ego-depletion. Upon arrival, participants were told that
they were going to participate in two different studies: a food
taste study and a visual perception study. Following the
procedures from Baumeister, Bratslavsky, Muraven, and Tice
(1998), two types of food were displayed on the table at which
participants were seated: chocolate chip cookies and red radish.
Participants were randomly assigned to taste one type of food,
and were asked to eat at least one unit (and as many as they liked)
of either cookie or radish, but not eat the other. Participants were
asked to take their time to taste the assigned food (average time
of food tasting about 5 min) before filling in a “Food Taste
Questionnaire”as a manipulation check for ego-depletion. In the
questionnaire they were asked to list down the instructions they
had remembered, as well as to report difficulty in following the
instructions and in resisting eating the other group's food on a
nine-point scale (1 = very easy, 9 = very difficult), then the
participants rated the taste and quality of the food they tasted.
Participants who ate radish but could not consume the cookies
were expected to form the ego-depleted group.
Participants then watched the butterfly slideshow as in
experiment 2. Three days later, they received and responded to a
questionnaire asking the slideshow duration judgment by e-
mail. This experiment therefore employed a 2 (slide transition:
repetitive vs. cyclical) × 2 (ego-depletion: depletion vs. no-
depletion) between-participants design.
Result and discussion
As a manipulation check for ego-depletion, we followed
Vohs and Schmeichel (2003) and compared the ease of
following instructions and difficulty in resisting the disallowed
food. The radish-eating group found it much more difficult to
follow instructions (M
= 3.41, M
= 1.67, F
(1, 59) = 9.19, pb.01) and found it more difficult to resist
eating the disallowed food (M
= 3.71, M
1.44, F(1, 59) = 20.48, pb.001). Hence our manipulation of
ego-depletion was successful.
The mean subjective duration judgments are plotted in Fig.
4, and were analyzed using a 2 (slide transition) ×2 (ego-
depletion) ANOVA. We found a significant two-way interaction
between the slide transition and ego-depletion (F(1, 60) = 5.54,
pb.05) and significant main effects of slide transition (F(1, 60)=
6.10, pb.05) and ego-depletion (F(1, 6 0) = 12.93, pb.01). The
main effect of slide transition was qualified by the two-way
interaction between the slide transition and ego-depletion. For
participants in the no-depletion condition, the repetitive
transition group (M=92) estimated a shorter duration than
the cyclical transition group (M=158; F(1, 2 9) = 7.32, pb.05).
However, under the depletion condition, both the cyclical
transition group (M=77) and the repetitive transition group
(M=75) estimated much shorter durations for the slideshow
and the mean difference between both groups was not
significant (F(1, 31) = 0.02, pN.9).
This experiment provided support for H3 by demonstrating
that when individuals had limited cognitive resources at
encoding stage, the number of memory markers taken and
recorded in our memory would be restricted, resulting in a lower
The objective of experiment 5 was three folds. Firstly, we
tried to replicate findings that support H1 with a different type
of stimuli; a video clip. Secondly, we tested H4; that the level of
sensitivity to the details of the external stimuli could moderate
the number of memory markers taken during the experience,
thereby impacting the delayed duration judgment. Thirdly, we
collected and coded unaided memory of the experience to
determine if it mediated the effects of our manipulations on
delayed duration judgments.
Participants, stimuli, procedure and measurement
One hundred and twenty-three university students partici-
pated in the study for cash compensation. The experimenter
Fig. 4. Duration as a function of ego-depletion and slide transition: Experiment 4.
513H.-K. Ahn et al. / Journal of Consumer Psychology 19 (2009) 508–516
explained that they would be watching a video clip and
randomly assigned them to one condition in a 2 (video clip:
sound effects vs. no sound effects) by 2 (timing of judgment:
immediate vs. delayed) by 2 (sensitivity level: high vs. low)
between-participants design. The video clip without sound
effects used in the experiment was a short documentary film in
which the Harry Potter movie production team explained the
making of the fireworks sequence in the movie. The video clip
with sound effects used the same video clip, but inserted various
sound effects whenever fireworks appearing in the video (15
times in total) to accentuate the memory for the fireworks.
Further, half of the participants were told to pay attention to
every single detail in the video as much as they could and to write
down suggestions on how the production could be improved (the
high sensitivity condition); while the other half were told to think
about the general theme of the video and to write down possible
titles for the video (the low sensitivity condition). As in
experiments 1 and 2, participants were either asked to estimate
the duration of the video right after they watched the video clip in
the immediate condition, or three days later via e-mail in the
delayed condition. In this questionnaire, participants were also
asked to list whatever they could recall about the video. The
open-ended responses were coded into individual memories, and
the number of fragments of memory (NUMMEM) was used as a
surrogate for the number of memory markers.
Result and discussion
The mean subjective duration judgments were analyzed
using a 2 (video clip) × 2 (sensitivity) × 2 (timing of judgment)
ANOVA. The mean and standard deviation scores of duration
judgment in each condition were listed in Table 1.
Firstly, sensitivity was shown not to have a significant main
effect (F(1, 115) = .95, pN.1), and hence we combined the high
sensitivity and low sensitivity groups for the analysis. We found
a significant two-way interaction of video clip by timing (F(1,
115) = 10.45, pb.01). The video clip with sound effects was
estimated to be shorter (M= 171) than the one without sound
effects when the judgment was made immediately after viewing
the video (M= 184); while the video clip with sound effects was
estimated to be longer (M= 335) than the one wi thout sound
effects when the judgment was made three days later (M= 223).
This result was essentially a replication of findings in
experiment 1 and 2 with a different type of stimuli. However,
the three-way interaction of 2 (video clip) × 2 (sensitivity) × 2
(timing) was not significant (F(1, 115) = 1.21, pN.2).
Second, when looking at the delayed duration judgment
only, we found that sensitivity had a moderating role on the
effect of video clip on duration judgment. When viewing the
video clip without sound effects, participants with a high
sensitivity were expected to encode more memory markers,
hence estimate a longer duration (M= 279) than parti cipants
with a low sensitivity (M= 174, F(1, 28) = 7.50, pb.05). When
viewing the video clip with sound effects, participants with a
low sensitivity might take as many memory markers (M= 309)
as participants with a high sensitivity, hence there was no
difference in the duration judgment (M= 359, F(1, 33) = 2.98,
Third, for the delayed conditions, there is a significant two-
way interaction of sensitivity by video clip on the number of items
recalled [NUMMEM] (F(1, 60) = 7.5 3, pb.01). Sensitivity
moderates the effect of video clip on recalled items. When
viewing the video clip without sound effects, participants with a
high sensitivity recalled more items (M= 7.90) from the video
than those with a low sensitivity (M=5.00); while when viewing
the video clip with sound effects, participants with a high
sensitivity recalled as many items (M=7.27) as those with a low
sensitivity (M=7.58). The effect of sensitivity and video clips on
recalled items is consistent with the effect of sensitivity and video
clips on duration judgment, providing a compelling evidence for
the underlying process of how individuals memorize time.
We further tested the number of recalled items from the video
as a mediator of the effect of video clip and sensitivity on
duration judgment. Four cases were omitted from the analysis
for missing data. Three regressions were conducted: (1) video
clip and sensitivity as predictors of duration judgment (F(2, 59)=
11.03, pb.001; (2) video clip and sensitivity as predictors of the
number of recalled items from the video (F(2, 59) = 8.83, pb.001)
and (3) video clip, sensitivity, and the number of recalled items
as predictors of duration judgment (F(3, 58) = 49.85 , pb.001).
The coefficient of video clip dropped from .43 (t(59)= 3.78,
pb.001) to .10 (t(58)= 1.22, p= .23) and the p-value became
insignificant when the number of recalled item is included in
the regression. Similarly, the coefficient of sensitivity dropped
from .26 (t(59)= 2.36, pb.05) to .13 (t(58)= 1.84, p=.07) and
the p-value became insignificant when the number of recalled
item is included in the regression. We then run the Sobel Test
of the significance of the indirect effect of the mediator
(MacKinnon, Warsi, and Dwyer, 1995; Sobel, 1982).The
Sobel statistics for sensitivity was computed as 1.50 (p=.07)
and the Sobel statistics for video clip was 3.56 (pb.001). The
Sobel Test showed the number of recalled item as a significant
Duration judgment (in seconds) as a function of timing judgment, video clip, and sensitivity.
Immediate judgment Delayed judgment
Sound effect video No sound effect video Sound effect video No sound effect video
High sensitivity Mean 158.13
SD 61.23 66.90 111.12 107.32
Low sensitivity Mean 186.67
SD 76.78 93.39 148.79 102.10
Note. Cells with a different superscript differ significantly from each other (pb.05).
514 H.-K. Ahn et al. / Journal of Consumer Psychology 19 (2009) 508–516
mediator of the effect of video clip on duration judgment, and
a marginally significant mediator of the effect of sensitivity on
This result demonstrated that sensitivity could moderate the
number of memory markers taken during the experience,
thereby impacting the delayed duration judgment. Combined
with the finding of experiment 4, we could argue that duration
judgment was heavily influenced by either increasing or
inhibiting the mental capacity during the encoding process,
thus providing more support to the underlying process of the
memory marker model.
General discussion and conclusions
In this article, we present a model of consumers' judgment of
the duration of past experiences. The human mind encodes
memory markers, stores them in appropriate memory bins, and
later retrieves them to reconstruct the experience. In rich
environments (relative to impoverished ones), when the mind
encodes a lot of memory markers in a given objective duration,
time passes by quickly during the experience and the duration
does not seem very long. However, when the experience is
recalled, the larger number of markers serves as a cue for a
longer duration judgment. We found evidence for this basic
finding across a field study and a series of experiments with
different stimuli. We showed that ego-depletion and cognitive
load result in shorter delayed duration judgment, and also
showed that making participants more sensitive to the
experience moderates the effects. These findings are consistent
with claims made by prior work. In particular, McGrath and
Kelly (1986) argued that humans can distort their sensory input
when they are experiencing time and hence experience a
different pattern from the objective or “true”time. In particular,
they assumed that there might be some kind of standard of
expected stimulus rate which may vary with situations and can
be subject to adaptation. In this case, individuals are able to
notice the stimuli as “many”or “few”based on the rate that they
have been adapted. Enriched intervals are remembered as if they
were longer, while empty intervals are remembered as if they
were shorter than they actually were (as measured by objective
clock time). This insight is consistent with our current findings.
The goal of our research was to study processes that affect the
manner in which people recall and judge one aspect of an
experience, its duration. Another facet relates to the intensity of
various elements of the experience. In our model, we
manipulated our experiences to be affectively unremarkable to
suit our assumption that each memory marker had the same
affective intensity as other memory markers. Indeed, the next
phase of a research program on how consumers remember
experiences should study the effect of intensity on all three
stages of the process.
Our model also has an interesting parallel in spatial distance
perception. For instance, Downs and Stea (1973) and Sadalla
and Staplin (1980) compared two otherwise identical routes,
with the difference that one of them had more intersections
(crossroads) and features (e.g., shops and signs) than the other
one. They found that the route with more intersections and
features is recalled as a longer route. This finding was similar to
what the memory marker model would predict —when
consumers reconstructed the route, the presence of more
intersections and features gave them the ability to encode
(and later retrieve) more memory markers, which contributes to
the judgment of greater length.
At the end of every holiday season, one often hears people
remark how short the previous year seemed to be. Our work
would suggest that this tendency is higher for people who have
routine jobs (e.g., librarian) than those whose job schedules are
variant (e.g., advertising team) or work locations are different
(e.g., plumber), because the former will have fewer memory
markers to code. To test this idea, we measured individuals'
perception of the duration of the past year as a function of the
variability of their job schedules and work locations (as
measured by ratings on nine-point scales). The ratings on
both job schedule and work location (median split) were used as
independent variables in a two-way ANOVA. The result
showed that both work location (F(1, 95) = 127.58, pb.001)
and job schedule (F(1, 95) = 152. 22, pb.001) have significant
effects on the perceived duration of the past year, controlling for
how eventful this year is comparing to other years (as measured
by a scale). The more routine the job (in terms of both location
and schedule), the more likely that individuals will feel time
pass quickly and surprised by it.
While the present research has provided some interesting
answers to how people remember duration of past experiences,
it has also opened up further questions for research. In our
experiment, we cleanly manipulated the nature of the stimulus
to make it rich or impoverished. How does this process work in
the real world? What happens when people have to deal with
memory of multiple experiences at the same time? Does the
human mind re-categorize memory markers over time? How do
emotions associated with the experience influence the three
stage research? Future research could investigate some of these
questions in further depth.
Antonides, G., Verhoef, C. P., & Aalst, M. (2002). Consumer perceptions of
telephone waiting times. Journal of Consumer Psychology,12, 193−202.
Ariely, D., Kahneman, D., & Loewenstein, G. (2000). Joint comment on when
does duration matter in judgment and decision making? Journal of
Experimental Psychology: General,129, 524−529.
Baumeister, R., Bratslavsky, E., Muraven, M., & Tice, D. M. (1998). Ego
depletion: Is the active self a limited resource? Journal of Personality and
Social Psychology,74, 1252−1265.
Block, R. A. (1990). Models of psychological time. In R. A. Block (Ed.),
Cognitive models of psychological time (pp. 1−30). Lawrence Erlbaum
Block, R. A., & Zakay, D. (1997). Prospective and retrospective duration
judgments: A meta-analytic review. Psychonomic Bulletin and Review,4,
Downs, R., & Stea, D. (Eds.). (1973). Image and environment AVC: Chicago.
Fredrickson, B., & Kahneman, D. (1993). Duration neglect in retrospective
evaluations of affective episodes. Journal of Personality and Social
515H.-K. Ahn et al. / Journal of Consumer Psychology 19 (2009) 508–516
Glicksohn, J. (2001). Temporal cognition and the phenomenology of time: A
multiplicative function for apparent duration. Consciousness and Cognition,
James, W. (1890). The principles of psychology (Vol. 1). New York: Holt.
Kahneman, D., Fredrickson, B. L., Schreiber, C. L., & Redelmeier, D. A. (1993).
When more pain is preferred to less: Adding a better end. Psychological
Kellaris, J., & Kent, R. (1992). The influence of music on consumers' temporal
perceptions: Does time fly when you're having fun? Journal of Consumer
Kundera, M. (1999). Immorality. New York: Harper Collins Publishers.
MacKinnon, D. P., Warsi, G., & Dwyer, J. H. (1995). A simulation
study of mediated effect measures. Multivariate Behavioral Research,30,
McGrath, J. E., & K elly, J. (1986). Time and human interaction: Toward a
social psychology of time. New York: The Guilford Press.
Redelmeier, D. A., & Kahneman, D. (1996). Patient's memories of painful
medical treatments: Real-time and retrospective evaluations of two
minimally invasive procedures. Pain,68,3−8.
Sadalla, E. K., & Staplin, L. J. (1980). The perception of traversed distance
intersections. Environment and Behavior,12, 167−182.
Schmitt, B. (2003). Customer experience management: A revolutionary
approach to connecting with your customers. Hoboken, New Jersey: John
Wiley & Sons.
Sobel, M. E. (1982). Asymptotic confidence intervals for indirect effects in
structural equation models. In S. Leinhart (Ed.), Sociological Methodology
(pp. 290−312). Jossey-Bass: San Francisco.
Stroud, J. M. (1967). The fine structure of psychological time. Annals of the
New York Academy of Sciences,138, 623−631.
Treisman, M. (1984). Temporal rhythms and cerebral rhythms. Annals of the
New York Academy of Sciences,423, 542−565.
Vohs, K. D., & Schmeichel, B. J. (2003). Self-regulation and the extended now:
Controlling the self alters the subjective experience of time. Journal of
Personality and Social Psychology,85, 217−230.
Wyer, R. S., & Srull, T. K. (1986). Human cognition and its social context.
Psychological Review,93, 322−359.
Wyer, R. S., & Srull, T. K. (1989). Memory and cognition in its social context.
Hillsdale, New Jersey: Lawrence Erlbaum Associates.
516 H.-K. Ahn et al. / Journal of Consumer Psychology 19 (2009) 508–516