the effects of cue familiarity on episodic memory, scene
TRANSCRIPT
The Effects of Cue Familiarity on Episodic Memory, Scene Construction, and Imagining the Future
by
Jessica Robin
A thesis submitted in conformity with the requirements for the degree of Master of Arts
Graduate Department of Psychology University of Toronto
© Copyright by Jessica Robin 2011
ii
The Effects of Cue Familiarity on Episodic Memory, Scene
Construction, and Imagining the Future
Jessica Robin
Master of Arts
Graduate Department of Psychology
University of Toronto
2011
Abstract
Recent research has revealed many similarities between episodic memory, scene
construction, and imagination of the future. It has been suggested that scene construction is the
common process underlying memory and imagination, but no study to date has directly
compared all three abilities. The present study compared retrieval time, ratings of detail and
vividness for episodic memories, remembered scenes and imagined future events cued by
landmarks of high and low familiarity. Memories, scenes, and imagined episodes based on a
more familiar landmark as a cue were more quickly retrieved, more detailed, and more vivid.
This study was the first to demonstrate the effects of frequent encounters with a cue on memory,
scene construction and imagination of the future. Additionally, consistent results across
conditions, as well as stronger effects in the scene construction condition, provide further
evidence of a possible interdependence of episodic memory, imagination of the future, and
scene construction.
iii
Acknowledgments
I wish to thank Roxana Florica for her help with the transcription and coding of the interviews,
and Marilyne Ziegler for her assistance in the design and programming of the experiment. I‟d
also like to thank the members of the Moscovitch lab for their help and advice, and the members
of my committee, Gordon Winocur and Cheryl Grady, for their thoughtful questions and
comments. Finally, I‟m very grateful to my supervisor, Morris Moscovitch, for his continued
guidance, support and insightful commentary.
iv
Table of Contents
Acknowledgments ......................................................................................................................... iii
Table of Contents ............................................................................................................................ iv
List of Figures ................................................................................................................................. vi
List of Appendices .........................................................................................................................vii
1 Introduction ................................................................................................................................. 1
2 Methods .................................................................................................................................... 12
2.1 Participants ........................................................................................................................ 12
2.2 Pre-study Questionnaire..................................................................................................... 12
2.3 Study Procedure ................................................................................................................. 13
2.3.1 Episodic Memory Condition ................................................................................. 14
2.3.2 Scene Construction Condition ............................................................................... 15
2.3.3 Imagination of the Future Condition ..................................................................... 15
2.3.4 Post-Study Interviews ............................................................................................ 16
3 Results ....................................................................................................................................... 18
3.1 Inter-rater Reliability ......................................................................................................... 18
3.2 Episodic Memory Condition ............................................................................................. 18
3.2.1 Retrieval Time ....................................................................................................... 18
3.2.2 Detail and Vividness Ratings ................................................................................ 19
3.2.3 Interview Details .................................................................................................... 20
3.2.4 Order Effects .......................................................................................................... 20
3.3 Scene Construction Condition ........................................................................................... 21
3.3.1 Retrieval Time ....................................................................................................... 21
3.3.2 Detail and Vividness Ratings ................................................................................ 21
3.3.3 Interview Details .................................................................................................... 21
v
3.4 Imagination of the Future Condition ................................................................................. 22
3.4.1 Retrieval Time ....................................................................................................... 22
3.4.2 Detail and Vividness Ratings ................................................................................ 23
3.4.3 Interview Details .................................................................................................... 24
3.5 Cross-Condition Comparisons ........................................................................................... 25
4 Discussion ................................................................................................................................. 26
References...................................................................................................................................... 34
Figures ........................................................................................................................................... 39
Appendices .................................................................................................................................... 45
vi
List of Figures
Figure 1. Representation of one trial in the episodic memory condition of the experiment. Order
of slides and duration is indicated in the upper left corner of each frame.
Figure 2. Retrieval Time. a) Mean retrieval time (in seconds) for memories, scenes, and
imagined events based on high and low familiarity landmarks. Error bars indicate the
standard error of the mean for each group. b) Mean retrieval time (in seconds) for
memories based on high and low familiarity landmarks, in the group of participants
below the median retrieval time, and the group above the median retrieval time. Error
bars indicate the standard error of the mean for each group.
Figure 3. Detail and Vividness Ratings. a) Mean ratings of detail (on a 1-5 scale) for memories,
scenes, and imagined events based on high and low familiarity landmarks. Error bars
indicate the standard error of the mean for each group. b) Mean ratings of vividness (on
a 1-5 scale) for memories, scenes, and imagined events based on high and low
familiarity landmarks. Error bars indicate the standard error of the mean for each
group.
Figure 4. Number of Interview Details. a) Mean number of details described per memory, scene,
and imagined event based on high and low familiarity landmarks. Error bars indicate
the standard error of the mean for each group. b) Mean number of details described per
imagined event based on high and low familiarity landmarks, in the group of
participants below the median number of details, and the group above the median
number of details. Error bars indicate the standard error of the mean for each group.
Figure 5. Magnitude of differences across high and low familiarity landmarks. a) Mean
magnitude of difference in detail ratings across high and low familiarity landmarks,
across memory, scene and imagination conditions. Error bars represent standard error
of the mean for each group. b) Mean magnitude of difference in vividness ratings
across high and low familiarity landmarks, across memory, scene and imagination
conditions. Error bars represent standard error of the mean for each group. c) Mean
magnitude of difference in number of interview details described across high and low
familiarity landmarks, across memory, scene and imagination conditions. Error bars
represent standard error of the mean for each group.
vii
List of Appendices
Appendix A. Full list of Toronto landmarks used in study.
Appendix B. Interview coding guide.
1
1 Introduction
Throughout any given day, many of our waking hours are spent thinking of a time other than the
present. From remembering where the car is parked, or envisioning what to cook for dinner, to a
certain song reminding us of a friend‟s party years ago, or a glance through the calendar sending
us day-dreaming about next year‟s vacation plans, our present is constantly filled with reminders
of the past and musings of the possible future. Memory for the past and imagination of the future
may seem like very different functions. One allows us to relive real experiences from our past,
recalling the sights, sounds and other sensory information that we once actually experienced. The
other allows us to fabricate novel or even impossible events that are often far removed from any
real experience. However, the ease with which we mentally slide backward or forward in time
seems to indicate otherwise. Indeed, research comparing memory of past experiences to the
imagination of future ones has repeatedly shown that these two capacities are actually highly
related and likely rely on the same brain networks to operate. Several explanatory hypotheses
have been advanced to account for these similarities. One such hypothesis suggests that the
ability to mentally construct a scene is the common core process in both episodic memory and
imagination (Hassabis & Maguire, 2007). The aim of the present study is to elucidate further the
similarities between episodic memory, imagination, and scene construction by comparing them
directly using very similar paradigms and cues. A second aim is to determine how episodic
memory, imagination and scene construction are mediated based on the amount of one‟s
experience with the cue that prompts the remembered or imagined scene or event.
Early research suggesting a link between memory and imagination came from work with brain-
damaged patients. Tulving (1985) observed that patient K.C., who suffered from severe amnesia
due to bilateral medial temporal lobe (MTL) damage, was also unable to envision future
experiences in his life. Similar observations were made about patients D.B. and H.M., who
suffered from both severe amnesia and impaired imagination of the future (Klein, Loftus, &
Kihlstrom, 2002; Buckner & Carroll, 2007). Recently, two additional patients with hippocampal
damage stemming from diverse etiologies have also been reported to have parallel deficits in
memory for the past and imagination of the future (Kwan, Carson, Addis, & Rosenbaum, 2010;
Andelman, Hoofien, Goldberg, Aizenstein, & Neufeld, 2010). In contrast, another group has
reported cases of patients with developmental amnesia who appear to be unimpaired on
2
imagination tasks (Cooper, Vargha-Khadem, Gadian, & Maguire, 2011; Maguire, Vargha-
Khadem, & Hassabis, 2010). Thus, while the majority of patient work suggests a link between
memory for the past and imagination of the future, there is also some evidence that these abilities
may be dissociable.
Following these studies, further research revealed that it was not only amnesic patients who
show impairments in memory and imagination tasks. Addis and her colleagues (2009a) reported
that patients with mild Alzheimer‟s disease generated fewer relevant details both when recalling
events from the past and when imagining possible future events. The numbers of details
produced in memories of the past and in imagined future events were correlated, and were not
attributable to either phonemic or semantic fluency. Similarly, another study reports nearly
identical findings for patients with amnesic Mild Cognitive Impairment (aMCI), a common
precursor to Alzheimer‟s disease (Gamboz et al., 2010). Both Alzheimer‟s disease and aMCI
have been associated with atrophy of the medial temporal lobes, and specifically, the
hippocampus, linking both of these findings to the previous studies of hippocampal amnesia
patients. Another clinical population, schizophrenia patients, has also been shown to have poorer
abilities in remembering past events and imagining future ones as compared to healthy controls
(D‟Argembeau, Raffard, & Van der Linden, 2008). Patients with schizophrenia generated fewer
specific memories of the past as well as fewer specific imagined future events, and their abilities
on these two tasks were again correlated with one another. Finally, studies of healthy older adults
have repeatedly demonstrated a parallel decline in detail-richness across remembered and
imagined events, associated with age (Addis, Wong, & Schacter, 2008; Addis, Musicaro, Pan, &
Schacter, 2010).
Numerous studies of healthy younger adults have revealed even more similarities across memory
for the past and imagination for the future. Both tend to be more detail-rich and be more vividly
experienced if the event in mind has a positive valence or occurs temporally closer to the present
(D‟Argembeau & Van der Linden, 2004). Individual differences in visual imagery or emotion
regulation tendencies were reflected in the amount of sensory detail and other experiential
qualities across both memories and imagined experiences (D‟Argembeau & Van der Linden,
2006). In addition, imagined events based on more recently experienced contexts were more
vivid and detailed than ones based on novel or remotely experienced contexts, suggesting that
imagined events are somehow dependent on memories of the past (Szpunar & McDermott,
3
2008). Finally, freely generated past and future events have been found to follow the same
temporal distribution pattern, and these same patterns were shown across young, middle-aged
and older adults (Spreng & Levine, 2006).
Evidence from neuroimaging studies has further supported the similarities between episodic
memory and imagination of the future by identifying a common network of brain areas involved
in both abilities (Okuda et al., 2003; Addis, Wong, & Schacter, 2007; Szpunar, Watson, &
McDermott, 2007; Botzung, Denkova, & Manning, 2008; D‟Argembeau, Xue, Lu, Van der
Linden, & Bechara, 2008; Addis, Pan, Vu, Laiser, & Schacter, 2009; Weiler, Suchan, & Daum,
2010; see Spreng, Mar, & Kim, 2009 for review). Though some differences exist between
studies, both in terms of methodologies and results, in general, the same core network of areas
has been repeatedly found to be active across memory and imagination tasks (Spreng et al.,
2009). This network is comprised of parts of the medial temporal lobes, medial parietal lobes,
and prefrontal cortex; more specifically, the bilateral hippocampus, parahippocampal gyri,
precuneus, posterior cingulate and retrosplenial cortices, tempoparietal junction, medial
prefrontal cortex and frontotemporal poles (Spreng et al., 2009).
In a recent study, Spreng & Grady (2010) highlighted the similarity of the memory/imagination
network to what‟s commonly referred to as the „default mode network‟, which is typically
engaged when individuals are at rest, or thinking freely. In addition, they used partial-least
squares analysis (PLS) to identify the high degree of overlap between brain regions engaged in
autobiographical memory, imagination of the future and theory of mind tasks, finding even
higher overlap between just the memory and imagination networks, particularly in the
hippocampus and regions along the frontal and parietal midline. In another study, the content of
the mental experience was held constant (a short walk in a familiar setting) but the time period
was varied between remembered past, and imagined past, present and future (Nyberg, Kim,
Habib, Levine, & Tulving, 2010). A very similar frontoparietal network was found to be active
for all of the tasks in the study, with an area in the left parietal cortex in particular activating for
all non-present time points. No hippocampal activation was observed. These results provide
additional evidence for the neural similarity of projecting oneself forward or backward in time,
and may indicate that the role of the hippocampus in memory and imagination is more related to
the content of the mental experience, since this study was unique in holding that constant across
tasks.
4
Taken together, these many similarities in terms of the phenomenology and the neural substrates
of episodic memory and imagination of the future make a strong case for the existence of a
common process or processes underlying these abilities. Several suggestions have been proposed
for what the common process(es) might be (for review, see Schacter, Addis, & Buckner, 2007).
Buckner and Carroll (2007) hypothesized that self-projection was the ability common not just to
memory and imagination, but also to navigation and theory of mind tasks. They defined self-
projection as the ability to shift one‟s perspective from the present, real-life situation to another
imagined alternative, and suggested that activity in the core network identified in neuroimaging
studies facilitates this ability.
Addis and Schacter advanced an alternative suggestion: the constructive episodic simulation
hypothesis (2008; Addis, et al., 2007; Schacter & Addis, 2009). This hypothesis states that
imagined future events are created by recombining elements from past episodic memories in
novel ways. According to this view, both memory and imagination are reconstructive processes,
but imagination requires more intensive construction activity since it involves combining more
disparate details into a new coherent representation. Three studies which all found increased
hippocampal activity during the construction of future events lend support to this hypothesis and
to the notion that the hippocampus is the locus of event construction (Addis et al., 2007; Addis &
Schacter, 2008; Addis, Cheng, Roberts, & Schacter, 2010).
A third way to explain the similarities between episodic memory and imagination has been
suggested in work by Hassabis, Kumaran, Vann, and Maguire (2007). Instead of focusing on
imagined personal events, they studied imagined scenes that have no involvement of the self, no
narrative structure and no specific temporal context. They found that patients with hippocampal
amnesia were impaired at imagining these scenes, and in particular, that what they could imagine
was lacking in spatial coherence. A related study identified a scene construction network in
healthy controls, consisting of many of the same areas as the network consistently found to be
involved in memory and imagination (Hassabis, Kumaran, & Maguire, 2007). These findings led
to a suggestion that the underlying process common to memory and imagination is the process of
scene construction, and that the hippocampus is crucial for this process (Hassabis & Maguire,
2007; Hassabis & Maguire, 2009).
5
A study with the amnesic patient K.C., whose inability to imagine his future served as the
catalyst for much of the research on memory and imagination, may lend further support to
Hassabis and Maguire‟s hypothesis. Despite extensive bilateral damage to the hippocampus,
K.C.‟s ability to navigate in the neighbourhood in which he grew up and lived in prior to his
brain damage is relatively preserved (Rosenbaum et al., 2000). However, his ability to identify
houses or other landmarks was impaired, and lacked detail. In another study, patient S.B., who
had medial temporal lobe damage due to Alzheimer‟s disease, was also unimpaired in navigation
but deficient in recognizing landmarks and conjuring visual imagery (Rosenbaum, Gao,
Richards, Black, & Moscovitch, 2005). Both of these cases demonstrate patients in whom both
episodic memory and detailed scene memory are impaired, yet spatial memory for navigation is
intact. If scene construction is a crucial component of episodic memory, it is not surprising that
both of these abilities are impaired in these patients, while navigation is preserved since it relies
on schematic representations, not detailed scenes (Rosenbaum et al., 2000).
Rosenbaum, Gilboa, Levine, Winocur, and Moscovitch (2009) also reported that K.C. was
impaired on a semantic narrative task involving the retelling of fairy tales or bible stories learned
long before he sustained brain damage. This task did not involve any self-projection and did not
draw on any past personal memories, and thus provides additional arguments against both the
self-projection and constructive episodic simulation hypotheses. Although the spatial coherence
of K.C.‟s narratives was not directly assessed, it was found that his narratives were significantly
lacking in detail and coherence, similar to his imagination of the future and memory of the past.
This study could therefore support either the scene construction hypothesis advanced by
Hassabis and Maguire, or the idea that a more general deficit in binding processes exists, which
is not limited just to scenes, as Rosenbaum et al. (2009) suggest.
Additional support for the idea that a deficit in the ability to mentally construct and represent
scenes is linked to deficits in both memory for one‟s past and imagination of one‟s future comes
from work with schizophrenia patients (Raffard, D‟Argembeau, Bayard, Boulenger, & Van der
Linden, 2010). Building on their previous research showing impairments in memory and
imagination in patients with schizophrenia (D‟Argembeau et al., 2008a), Raffard et al. also
demonstrated that these patients were impaired on the same scene construction task found to be
impaired in hippocampal amnesic patients (Hassabis et al., 2007). Similar to the amnesic
6
patients, schizophrenia patients also produced less detailed scenes that were lower in measures of
spatial coherence than matched controls.
Several recent studies have also touched on the possibility of a link between the mental
representation of scenes and autobiographical memory and imagination. In a study by Gaesser,
Sacchetti, Addis and Schacter (2011), the authors replicated previous findings that older adults
provide fewer internal (episodic) details when asked to remember events from their past or
imagine possible future events. They also found the same pattern when older adults were asked
simply to describe a picture of a complex scene; once again the older adults provided fewer
relevant details, even in this non-mnemonic task. Regression analyses revealed that performance
on the picture description task significantly predicted performance on the memory and
imagination tasks. This study suggests that some non-mnemonic common mechanism underlies
performance on all three of these tasks, which could be related to the ability to perceive and
mentally represent complex scenes, since this ability is involved in all three tasks. However,
another recent study employed very similar methodology, but tested amnesic patients with
medial temporal lobe lesions instead of older adults (Race, Keane, & Verfaellie, 2011). In this
group, they found a parallel deficit across memory and imagination of the future tasks, but in
contrast to the Gaesser et al. (2011) study, no impairment in the picture description task. The
authors interpreted this finding to mean that while patients with amnesia are impaired in both
remembering the past and imagining the future, this is not due to a general inability to construct
narratives. In this study, the patients were not impaired at describing complex scenes while they
were present visually, but it was not tested if their ability to do so from memory was also
impaired, or if this impairment was correlated with their deficits in both episodic memory and
imagination. Together these studies demonstrate the recent interest in the relationship between
memory, imagination, and scenes, as well as the need for further study into the underlying
processes or mechanisms mediating this relationship.
In another recent study, participants were asked to imagine events at various time points in the
future and to rate them on a wide range on phenomenological qualities (Arnold, McDermott, &
Szpunar, 2011). It is already known that events further away either in the past or the future tend
to be remembered or imagined less vividly (D‟Argembeau & Van der Linden, 2004), and this
study revealed that the only quality predicting this decline in vividness was the „clarity of
location‟, or spatial coherence of the event. This suggested that the clarity of context or location
7
of a memory or imagined event was the primary determinant of its vividness, indicating the
importance of mental representations of scenes for both memory and imagination. A second
study from the same group showed that certain parts of the common neural network for
memories and imagined events were only activated if the imagined or remembered events were
set in a familiar context (Szpunar, Chan, & McDermott, 2009). If the context was unfamiliar,
these parts of the network, specifically the posterior cingulate cortex, parahippocampal cortex
and the superior occipital gyrus, were much less active. These neural areas have been implicated
in tasks involving autobiographical memory, mental navigation and representations of scenes,
leading the authors to propose that the activity observed during memory and imagination tasks in
familiar contexts could be related to the representation of these familiar contexts, or scenes,
which again implies that scenes are a key underlying component of both memory and
imagination.
In a related vein, there is considerable research showing that the hippocampus, which is known
to be an essential brain area for both episodic memory and imagination of the future (see above),
is also crucially involved in the representation of scenes. Although this does not necessarily
signify that these processes utilize the hippocampus in the same way, it could be an indication of
yet another link between these abilities. In a study comparing episodic judgments and spatial
judgments based on geographic landmarks, it was found that both produced significant activity in
the network common to episodic memory and imagination of the future, in particular in a
bilateral middle hippocampal area (Hirshhorn, Grady, Rosenbaum, Winocur, & Moscovitch,
submitted). Hirshhorn et al. speculated that this overlapping activity could be due to the fact that
both types of judgments were engaging scene construction processes and that these processes are
mediated bilaterally by an area in the middle hippocampus. Furthermore, Hirshhorn, Newman
and Moscovitch (2010) reported that the number of details remembered from episode-like spatial
memories correlated positively with tests of hippocampal function in a group of older adults,
whereas measures of schematic or map-like knowledge had no correlation. Both of these studies
support the notion that highly detailed spatial memories, or scenes, are dependent on the
hippocampus.
Adding further support is a recent study employing the relatively novel technique of multivariate
pattern analysis (MVPA) in conjunction with high-resolution fMRI to determine the areas of the
brain involved in the representations of scenes (Bonnici et al., 2011). They found that the most
8
detailed representations of scenes were present in the hippocampus, although other areas of the
medial temporal lobes, like the entorhinal cortex and the parahippocampal gyrus, were involved
in scene processing as well. Finally, two studies comparing episodic and semantic spatial and
non-spatial memories both found that the hippocampus was activated both for spatial memories
and episodic memories, and that episodic-spatial memories in particular led to the highest levels
of hippocampal activity (Ryan, Lin, Ketcham, & Nadel, 2010; Hoscheidt, Nadel, Payne, & Ryan,
2010).
Thus, although a vast body of research seems to indicate a link between the representations of
scenes, episodic memory and imagination of the future, the nature of this link is still not clear
and few studies have compared all three abilities in a single paradigm. The present study aims to
do so by comparing behavioural measures such as reaction time and ratings of detail and
vividness across an episodic memory task, a scene construction task and a future imagination
task while keeping all other aspects of the methodology constant. If the same patterns of results
are observed across all three tasks, this will provide support for common processes underlying
the three abilities. If these effects are greater in the scene construction condition, this could
indicate that scene construction is a key component process on which both episodic memory and
imagination rely. Alternatively, if the scene construction task shows patterns of results differing
from both the episodic memory task and the imagination task, this would provide an argument
against Hassabis and Maguire‟s scene construction hypothesis, and perhaps indicate that
although scene construction shares some similarities with memory and imagination, it is not a
necessary component process of either.
The second purpose of the study is to examine the effects of the familiarity of a cue on memory,
imagination and scene construction. The effects of many factors on the phenomenology of
memories and imagined experiences have been observed, including emotional valence, temporal
distance, and novelty of context (D‟Argembeau & Van der Linden, 2004; Szpunar &
McDermott, 2008). In Szpunar and McDermott‟s study (2008), they observed that cuing
participants with more recently experienced contexts elicited more detailed and more vivid
imagined events than ones based on remotely experienced or never experienced contexts. In a
later study, Arnold, McDermott and Szpunar (2011) found that when asked to imagine
something in the near (rather than far) future, participants were more likely to place the
imaginary event in a more familiar location or context. They also found that when asked to
9
imagine events in familiar versus unfamiliar locations, the events set in familiar locations were
imagined more clearly, and easily.
However, when examining the effects of familiarity of context, most previous studies have
simply compared „familiar‟ with „unfamiliar‟. It is not surprising that it is easier to imagine an
event in a well-known context versus something never before experienced. No previous study
has looked instead at the effects of a highly familiar cue versus a cue that is still familiar, but to a
lesser degree. If someone has dozens of memories associated with a given cue, is it easier or
more difficult to retrieve a single one of them than if they had only one or two memories for that
cue? Once retrieved, is that individual memory also more vivid and richer in detail, or no
different? Furthermore, do these effects persist when imagining a possible future event or
conjuring a scene based on that same cue? This study aims to answer these questions, thereby
elucidating the effect of the level of familiarity with a cue on memory, imagination and scene
construction. To assess these effects, we collected information about how phenomenological
factors, such as vividness or detail-richness, vary based on the level of the familiarity of the cue,
and measured retrieval time to determine how varying familiarity levels affect the ease of
retrieval of memories, scenes, and imaginary events.
There has been little research to date regarding how having multiple similar or overlapping
episodic memories affects retrieval of one of those memories. However, a similar phenomenon
has been studied in the context of word recognition for decades. Scarborough, Cortese &
Scarborough (1977) were among the first to report the phenomenon that high frequency words
(i.e. ones that appear most often in everyday speech) were recognized more quickly than low
frequency words in lexical decision tasks. This observation sparked extensive further study of
how and why this effect occurs and how it relates to the organization of the mental lexicon
(Morrison & Ellis, 1995; Monsell, Doyle, & Haggard, 1989; Malmberg, Steyvers, Stephens, &
Shiffrin, 2002). It is possible that these effects also extend to memories and scenes, meaning that
more frequently encountered cues would lead to faster and easier recall of memories associated
with those cues.
An opposite prediction comes from research on what has been dubbed the „fan effect‟
(Anderson, 1974).This effect occurs when many facts are known about a certain concept, and
consequently, each individual fact or piece of information becomes harder to retrieve (Anderson
10
& Reder, 1999). This effect has been shown in many contexts, including face recognition and
tests of semantic knowledge. The fan effect research would predict that as a cue is associated
with more memories, each individual memory takes longer, and more effort, to access. Although
these two effects both present plausible predictions of how the familiarity of a cue may affect
memory retrieval, neither provides a very analogous situation to the recall of complex episodic
memories concerning real or imagined autobiographical events. Both word frequency effects and
the fan effect have been observed in the context of recognition memory, not recall, and both are
found in the context of simple, more semantic-like stimuli, such as words, facts or faces.
For these reasons, it is of interest to determine familiarity effects on retrieval in the context of
personal, episodic memories, complex scenes and imaginary future events. As in the study of
word recognition, the existence or non-existence of familiarity effects may be informative
regarding the organization and storage of episodic memories and spatial scene information. If
multiple exposures to a given landmark result in faster retrieval of a memory or scene, this could
mean that increased experience with a cue acts in a cumulative fashion, resulting in stronger and
more easily accessible memories. Alternatively, multiple exposures to a cue could result in an
increased number of unique memory traces that interfere with one another, causing a longer
search and therefore less accessible memories.
Another phenomenon of interest is the effect of increased familiarity with a cue on the detail
richness and vividness of the memories, scenes, and imagined events. According to the Construal
Level Theory, as the “psychological distance” of something increases, its representation, or
construal, becomes more abstract (Liberman & Trope, 2008). Evidence for this effect has been
found in terms of actual spatial distance, temporal distance, and social distance. For example,
activities imagined in the more distant future tend to be described in more abstract, less detailed
terms (Liberman & Trope, 2008). These effects have also been shown in other research, such as
Szpunar and McDermott‟s (2008) finding that imagined events based on more recently
experienced contexts tend to be more vivid and detail-rich, while more remotely experienced
contexts led to less detailed imaginary events. In this study, we sought to determine if familiarity
represents another facet of psychological distance, in that more familiar cues are somehow
„closer‟ conceptually than less familiar cues. If so, it would follow that memories, scenes, and
imaginary events that are based on more familiar cues should be more detailed, whereas those
based on less familiar cues should be more abstract and general in nature. Additionally, if
11
increased exposure to a certain cue leads to more detailed and vivid memories and imagined
events, then perhaps the reported recency of context effect (Szpunar & McDermott, 2008) is just
a product of higher exposure to the more recently experienced cues, and, therefore, can be
explained better by the familiarity of the context, not its recency.
To summarize, the present study seeks to determine what the effects are of varying familiarity
levels of cues across the related tasks of episodic memory, spatial scene construction and
imagination of future events. In order to test this, we cued participants with the names of public
landmarks in the city of Toronto of varying levels of familiarity to them. We then measured the
length of time it took them either to recall a past memory related to that landmark, to picture the
scene including that landmark or to imagine a future event that could potentially occur at, or
around, that landmark. They rated these memories, scenes and imaginary events for the level of
detail and vividness, and noted when they occurred or when they were imagined to occur.
Finally, participants also were asked to describe some of these remembered and imagined events
and scenes out loud and were recorded while doing so, in order to allow for an additional
objective assessment of detail-richness based on their descriptions.
In line with Construal Level Theory (Liberman & Trope, 2008), we predicted that increased
familiarity with a landmark via multiple past experiences would generate more vivid and detailed
memories, scenes, and imagined events involving that landmark due to a richer and more
complex memory representation. Participants presumably draw on these more numerous past
memories of the landmark when either reconstructing an old memory, conjuring the scene or
constructing a new event, and hence these representations will be more detail-rich and more
vivid than constructions based on less familiar landmarks. We also predicted that these overall
familiarity effects would exist beyond any effects of recency of visiting the landmark in
question. In line with the Hassabis and Maguire‟s (2007) hypothesis that scene construction is an
underlying process linking episodic memory and imagination, and the body of research showing
links between the three tasks, we expected that participants would show the same patterns of
amount of detail and level of vividness in response to cue familiarity across the episodic
memory, imagination and scene construction tasks.
In terms of ease of retrieval, we expected that the greater the familiarity of the cues the faster the
retrieval times in scene construction and imagination of future events since in these tasks there is
12
no need to choose a single representation from memory, and in fact, higher familiarity with a
landmark may help to bring the scene to mind faster or to conjure a novel event more easily.
However, we predicted that the retrieval time results for the episodic memory task would
perhaps not follow the same pattern. It was possible that when only a single memory had to be
recalled, it would be easier to do so if interference from other memories was minimized.
Following this, we thought that it may be the case that memories associated with landmarks that
are visited less often would be easier to retrieve because they are drawn from a smaller pool,
thereby reducing the interference from competing memories. On the other hand, it was also a
possibility that more familiar cues led to all the memories associated with those cues being more
readily accessible, and thus faster and easier to retrieve. Due to these conflicting predictions and
very little previous research, we did not have a specific prediction in terms of the retrieval time
results for the episodic memory condition.
2 Methods
2.1 Participants
56 healthy young adults (16 male, mean age = 21.00; SD = 2.94, range = 18-31) participated in
the experiment either for course credit or for monetary compensation (10$/hour). All participants
stated that they frequently visit the downtown area of Toronto (at least several times per month),
and had lived in Toronto for at least one year (mean years lived in Toronto = 11.07; SD = 7.54),
ensuring that they had a variety of old and new memories involving the landmarks featured in the
study. Participants had completed an average of 14.80 years of formal education (SD = 2.16),
were all native or fluent speakers of English, had normal or corrected-to-normal vision and
hearing, and no history of neurological illness or injury. All participants provided informed
consent prior to participating in the experiment, in accordance with the University of Toronto
Office of Research Ethics.
2.2 Pre-study Questionnaire
At least 24 hours prior to the study, participants completed an online questionnaire to assess their
familiarity with a variety of well-known Toronto buildings and landmarks, such as the CN
Tower or Union Station (full list of landmarks used in Appendix 1). The questionnaire provided
a list of 112 landmarks located mostly in downtown Toronto, 60 from the original Toronto
13
Public Places Test (Rosenbaum et al., 2004), as well as 52 additional landmarks, and asked
participants to estimate the number of times they‟d visited each of the landmarks (response
options: never, 1-2 times, 3-5 times, 6-10 times, more than 10 times). Participants were informed
that if they were unsure of whether they‟d visited the landmark, or were unfamiliar with the
name, to select „never‟. In addition, it was noted that the definition of „visiting‟ should include
walking by the landmark or viewing it from the exterior only, as well as entering the building or
location in question.
For the purposes of this study, landmarks visited between one and five times were considered
„low familiarity‟ and landmarks that have been visited more than ten times were considered „high
familiarity‟. Only these two categories of landmarks were used as stimuli for the study in order
to create a significant difference in the familiarity of the landmarks, while still ensuring that the
participants had visited all the landmarks at least once. Based on each participant‟s questionnaire
responses, a set of at least twenty „low familiarity‟ landmarks and at least twenty „high
familiarity‟ landmarks was selected, and used as stimuli in their unique version of the
experiment. Any participant who failed to classify at least twenty landmarks in each of these
categories was not eligible to participate in the study.
2.3 Study Procedure
The experiment was designed and run using E-Prime 2.0 software. It included three conditions:
episodic memory, scene construction and imagination of the future. Every participant completed
at least two of the three conditions, and participants with enough eligible landmarks in the high
and low familiarity categories completed all three conditions (N = 8). Each condition consisted
of 20 trials (10 using high familiarity landmarks as cues, and 10 using low familiarity landmarks
as cues), for a total of either 40 or 60 trials in the study, depending on the number of conditions
completed. Each landmark was randomly assigned to one of the conditions, and was only used
once in the study. The study was blocked by condition in order to minimize any confusion
between the tasks, and the order of the conditions was randomized and counterbalanced across
participants to eliminate any order effects. Due to the increased number of dropped subjects in
the memory condition, it was not possible to counterbalance the order of conditions for
participants in the memory condition, but see Results section for discussion of effects of
condition order. Before starting the study, participants were shown an example of each trial type
14
by the experimenter, and then completed two practice trials for each condition to ensure that they
understood the tasks involved in the study.
2.3.1 Episodic Memory Condition
Fifty-five participants completed the episodic memory condition. Two participants were dropped
due to failure to follow instructions, 15 were dropped due to inability to produce memories in
more than half of the low familiarity trials, and three participants were dropped due to very slow
reaction times (more than 2 SDs higher than the mean), resulting in 35 participants remaining in
the experiment. In the memory condition, participants were asked to recall past personal episodes
occurring at or around ten high familiarity landmarks and ten low familiarity landmarks
randomly selected from their pre-study questionnaires. Prior to starting the trials, participants
were instructed that they should recall events both specific in time and in place (i.e. no longer
than one day in duration and occurring in close proximity to the landmark in question). In
addition, participants were asked to recall only events that had occurred at least one month prior
to the study.
During the study, participants were seated in a quiet room facing a computer screen. At the start
of each trial, a white screen displayed the prompt, “Recall an event involving…” Two seconds
later, the name of a landmark appeared on the screen and the participant was asked to press the
spacebar as soon as a memory involving that landmark came to mind, providing a measure of
retrieval time. The landmark remained on the screen for a maximum of ten seconds, during
which a beeping noise played, to remind the participants to press the button as soon as they
thought of a relevant memory. If the participant failed to retrieve a memory by ten seconds, the
trial proceeded automatically, and was not included in the analyses. Following this retrieval
phase was a mental elaboration phase, in which the instructions, “recall and replay the event in as
much detail as possible”, were presented for 20 seconds, as participants attempted to remember
as many details as possible about the selected memory. After 20 seconds, the participant was
presented with three rating scales and was asked to assess the memory in terms of amount of
detail (1 – not very detailed, to 5 – very detailed; or 0 – no event); vividness (1 – not very vivid,
to 5 – very vivid; or 0 – no event); and length of time since the event actually occurred (0 – no
event, <1 month, 1-6 months, 6-12 months, >1 year, >5 years). Between each trial there was a
three-second fixation cross. Participants performed a total of 20 trials in this condition, with the
15
10 high familiarity and 10 low familiarity landmarks appearing in a random order. The structure
of one trial from the episodic memory condition is shown in Figure 1.
2.3.2 Scene Construction Condition
Thirty-three participants completed the scene construction condition. Two participants were
dropped due to failure to follow instructions, four were dropped due to inability to produce
scenes in more than half of the low familiarity trials, and three participants were dropped due to
very slow reaction times (more than 2 SDs higher than the mean), leaving only 24 participants.
The scene construction condition followed the same format as the episodic memory condition,
except that the task was to recall and picture visual scenes, not episodes. Participants were
instructed to picture the landmark named on the screen, and the area surrounding it, in as much
detail as possible. They were also instructed to avoid recalling any specific events or people that
they associated with that landmark, thus focusing on atemporal and impersonal representations
of the landmark rather than specific episodes. Each trial began with a prompt instructing
participants to “picture the scene around…”, and then the name of a landmark appeared on the
screen. Participants were instructed to press the space bar as soon as an image of the scene was
in mind, and a beep was played to remind them to perform this task. Following the button press,
or a maximum of ten seconds, there was once again a 20-second elaboration phase in which they
were asked to visualize the scene and to conjure as many details as possible. At the end of the 20
seconds, participants were asked to rate the details and vividness of the scene on the same rating
scales as in the episodic condition, and then to indicate the most recent time that they visited that
landmark (Never, <1 month ago, 1-6 months ago, 6-12 months ago, >1 year ago, >5 years ago).
Again, there was a total of 20 trials in the scene construction condition, with 10 landmarks of
each high and low familiarity occurring in a random order, with no landmarks repeated from the
other conditions, and a three-second fixation cross separating each trial.
2.3.3 Imagination of the Future Condition
Thirty-two participants completed the imagination condition. Five participants were dropped due
to inability to produce an imagined event in more than half of the low familiarity trials, and one
participant was dropped due to very slow reaction times (more than 2 SDs higher than the mean),
leaving 26 participants. In the imagination of the future condition, participants were asked to
conjure a plausible future event involving themselves and the landmark presented on the screen.
16
As in the episodic memory condition, they were instructed to imagine events that are specific in
time and place, and were asked to conjure events distinct from any past memories involving the
landmark in question. In addition, it was noted that each imagined event should differ in content
from one another, and not simply be the same event occurring in different settings.
As with the other conditions, an initial prompt appeared on the screen (“imagine a future event
involving…”), followed by the name of either a high or low familiarity landmark. Participants
were asked to press the spacebar once they had an imaginary future event in mind. Following the
key-press or a maximum of ten seconds, there was a 20-second elaboration phase in which
participants were asked to play the imagined event in their mind and conjure as many details as
possible. Participants were then asked to rate the imagined event for the amount of detail and
vividness on the same scales as the other two conditions, and to indicate how far in the future the
imagined event took place (No event, <1 month, 1-6 months, 6-12 months, >1 year, >5 years).
Finally, they were asked to judge how similar the imagined event was to a past memory on a
rating scale ranging from 1 – completely different, to 5 – extremely similar.
In all three conditions, if participants failed to press the spacebar indicating a memory, scene or
imagined event was in mind, or chose „0 – no event‟ for any of the rating scales, that trial was
discarded from the analysis.
2.3.4 Post-Study Interviews
Following the computer trials, a short interview was performed with each participant, in order to
obtain an objective measure of detail in conjunction with the participants‟ subjective ratings. In
the interview, two or three high familiarity and two or three low familiarity landmarks were
selected from each condition and participants were asked to describe in detail the memory, scene
or imagined event that they conjured based on that landmark. The interview techniques were
based on the Autobiographical Interview (Levine, Svoboda, Hay, Winocur, & Moscovitch,
2002), in which participants were first asked to freely recall and describe the scene or event,
followed by some general probing (e.g. “are there any other details that come to mind?”). No
specific probing regarding particular types of details was performed. The participants were asked
to describe the scenes and events in as much detail as possible, and were advised that they could
opt to skip a certain landmark if they had failed to conjure a scene, memory or imaginary event
based on that landmark during the experiment, or did not wish to describe the associated event or
17
scene for any reason. The difference between the tasks was emphasized prior to the study and
again in the interviews. Participants were encouraged to focus on the non-spatial content of the
events for the memories and imagined experiences, and on visuospatial information in the scene
construction task, in order to minimize overlap or confusion between conditions.
The interviews were recorded using a digital voice recorder, and the sound files were transferred
to a computer and transcribed by a research assistant, and later verified by a second transcriber.
Transcribed interviews were then scored for the number of relevant details in each memory,
imagined event, or scene. For memories and imagined events, detail scoring was based on
guidelines from the Autobiographical Interview scoring manual, where relevant (or „internal‟)
details are defined as those that are directly related to the event being recounted, whereas
external details consisting of semantic or other extraneous information were not counted (Levine
et al., 2002). Following this procedure, the main event in each description was identified, and
any piece of information relating to the event itself, actions that occurred, the time, the place, the
people involved, sensory perceptions, thoughts or feelings felt or expressed at the time were all
counted as details. Unrelated events, general background or semantic information, reflections or
judgments of the memories or future events, and repetitions or similar statements were not
counted. Imagined future events were coded according to the same guidelines as memories,
except that uncertain statements using terms such as „probably‟ or „hopefully‟ were taken as
factual statements, due to the fact that people tend to describe imagined events in more uncertain
terms than actual memories.
The number of details in each scene description was also coded, according to separate guidelines.
For scenes, only visual or spatial information about the landmark or its surrounding area was
considered as a relevant detail. Descriptions of the building itself, colours, textures, placement of
windows, signs or doors, and similar descriptions of the area or buildings surrounding the
landmark were counted as details. Event-specific information such as the weather, the presence
of people, or any actions or events was not included since it is not part of the visual-spatial
representation of the scene. In addition, general knowledge or other semantic information about
the scene was not counted as a detail. For full coding guidelines and examples, see Appendix 2.
The number of relevant details was counted for each interview while the coder remained blind to
whether the landmark was of high or low familiarity to the participant. A randomly selected
18
20% of the interviews were additionally coded by a second coder to assess reliability and
consistency of coding methods.
Finally, for the interviews in the scene condition, it was also possible to assess the accuracy of
the interviews since participants were describing static, real-life landmarks. By comparing the
participants‟ descriptions of the scenes to online images and Google Maps „street views‟ of the
areas, it was possible to determine which details mentioned were accurate and which were not. If
certain descriptions focused on the interior of the building, or provided other unverifiable details,
these interviews were excluded from this analysis, resulting in the inclusion of 82% (103 of 125)
of the original scene interviews in the accuracy analysis.
3 Results
3.1 Inter-rater Reliability
Correlations were computed between the numbers of details counted for the ten subjects‟
interviews that were coded by both of the raters. There was high agreement between the raters,
with an overall correlation of r = 0.86 (correlations of r = 0.80, r = 0.89, and r = 0.92, for
memory, scene, and imagination conditions, respectively), verifying the consistency of the
coding methodology used.
3.2 Episodic Memory Condition
3.2.1 Retrieval Time
As shown in Figure 2a, in the episodic memory condition, there was a trend for participants to
take less time to retrieve a memory based on a high familiarity landmark than a low familiarity
landmark. A paired t-test revealed that this trend did not reach significance (t(34) = 1.671, p =
0.10). However, a median-split of the retrieval time data revealed very distinct trends across
participants who tended to be faster on this task and those who were in the slower half of
participants. Participants whose retrieval times were below the median showed a significant
difference between mean retrieval times for memories based on high familiarity versus low
familiarity landmarks, with high familiarity landmarks allowing for significantly faster memory
retrieval (t(16) = 3.244, p = 0.005; see Figure 2b). Participants with retrieval times above the
median, however, showed no significant difference in retrieval time across the high familiarity
19
and low familiarity landmarks (t(16) = 0.107, p = 0.92). Further examination of the data revealed
that if the twenty-two participants (63% of the total sample) with mean retrieval times under
three seconds were considered (overall mean RTs = 2.21 and 2.46 seconds, for high and low
familiarity landmarks, respectively), the difference in memory retrieval times between high and
low familiarity landmarks was even more robust (t(21) = 3.651, p = 0.001). In fact, the disparate
results shown by the median-split are likely due to a consistent reverse trend (i.e. slower retrieval
times for memories based on high familiarity landmarks) shown only by the participants with the
slowest reaction times. In just the five participants with the slowest reaction times, this reverse
effect exhibits a trend towards significance, despite the very low number of participants in the
sample (t(4) = 2.290, p = 0.084).
Importantly, it was found that only the reaction times differed between the two median-split
groups and that there were no differences between these groups in terms of any of the other
effects observed in the study. Therefore, the total participant sample was used for all subsequent
analyses.
3.2.2 Detail and Vividness Ratings
Detail and vividness ratings made by the participants were also significantly different for
memories based on high familiarity versus low familiarity landmarks. As shown in Figure 3a and
b, a memory based on a highly familiar landmark tended to be rated as significantly more
detailed and more vivid than a memory based on a less familiar landmark (t(34) = 5.499, p < 0.001
for detail ratings; t(34) = 4.959, p < 0.001 for vividness ratings). Memories based on highly
familiar landmarks also tended to be related to events that had occurred more recently than
memories based on less familiar landmarks (t(34) = 3.611, p = 0.001). However, a regression
analysis revealed that the familiarity of the landmark was the most significant factor in predicting
the level of detail of the memory, even when recency was also included as a predictor in the
regression (R2 = 0.260, F(2,67) = 11.780, p < 0.001; p = 0.001 for familiarity, p = 0.051 for
recency). The same results were also found when predicting the level of vividness based on
familiarity and recency (R2 = 0.232, F(2,67) = 10.092, p < 0.001; p = 0.003 for familiarity, p =
0.040 for recency), demonstrating the importance of cue familiarity on the phenomenology of
memories.
20
3.2.3 Interview Details
The number of details described in the interview portion of the experiment was also significantly
higher for memories based on highly familiar landmarks, thus showing the same pattern as the
subjective detail ratings (t(34) = 4.564, p < 0.001; see Figure 4a). The number of details described
in the interviews was significantly correlated with the subjective ratings of detail, indicating
good agreement between subjective and objective measures of detail (r = 0.386, p = 0.001). Also
similar to the detail ratings, a regression analysis revealed that the familiarity of the landmark
was the most important predictor of the number of details described for the memories, whereas
recency was revealed not to be a significant predictor of detail in the interviews (R2 = 0.113,
F(2,67) = 4.271, p = 0.018; p = 0.006 for familiarity, p = 0.134 for recency).
The interviews were also examined in terms of event-related versus spatial or scene-related
details. It was found that the participants focused on describing the event-based content of the
memories, and included very few general scene-based details in the descriptions of the
memories. This confirmed that participants successfully differentiated between the differing
tasks of the study and were not simply reporting similar descriptions across all three conditions.
Note that if the alpha level is set at 0.0055 following the Bonferonni correction for multiple t-
tests (to maintain family-wise error rate of 0.05), the significance of the above results does not
change.
3.2.4 Order Effects
As mentioned in the Methods section, due to the high number of participants who did not
successfully complete enough trials in the memory condition, the orders of conditions in the final
group of participants in the memory condition were not perfectly counterbalanced. To investigate
the effects of this, several mixed ANOVAs were performed with condition order as the between
subjects variable, and landmark familiarity as the within-subjects variable. These ANOVAs
revealed no effect of condition order, or any interaction between condition order and familiarity
of the landmark for any of the dependent variables (retrieval time, detail ratings, vividness
ratings or number of interview details; all p‟s > 0.27). Whether the other condition was
imagination of the future or scene construction did have an effect on retrieval time (F(1,27) =
12.296, p = 0.02), but not on any other measures (all p‟s > 0.34). Post-hoc comparisons revealed
21
that participants tended to conjure memories faster if they also completed the scene condition
rather than the imagination condition (t(27) = 2.996, p = 0.006 for RTs based on high familiarity
landmarks, t(27) = 3.562, p = 0.001 for RTs based on low familiarity landmarks). Importantly
though, there was no interaction between this effect and the effect of landmark familiarity, so we
have no reason to believe that the slight imbalance in conditions impacted any results of interest
in this study.
3.3 Scene Construction Condition
3.3.1 Retrieval Time
As is evident in Figure 2a, participants had a faster mean retrieval time for scenes based on
highly familiar landmarks as compared to scenes based on less familiar landmarks. A paired t-
test confirmed the significance of this difference (t(23) = 3.591, p = 0.002), showing that if the
participant was more familiar with the landmark, it took less time to bring the image of the scene
to mind. Unlike the memory condition, there was no difference in effects based on overall speed
of retrieval, and therefore no median-split analyses were performed.
3.3.2 Detail and Vividness Ratings
Figure 3a and b demonstrate a clear difference between subjective ratings of detail and vividness
for scenes based on high and low familiarity landmarks. High familiarity landmarks led to scenes
that were rated as more detailed and more vivid, as confirmed by paired t-tests (t(23) = 12.497, p <
0.001 for detail ratings; t(23) = 14.174, p < 0.001 for vividness ratings). Again, the recency of the
last visit to the landmark also differed significantly across high and low familiarity landmarks,
with more familiar landmarks having been visited more recently (t(23) = 7.918, p < 0.001).
Regression analyses revealed that both familiarity and recency were significant predictors of
ratings of detail and vividness, with familiarity as a slightly more significant factor (detail
ratings: R2 = 0.727, F(2,45) = 60.022, p < 0.001; p < 0.001 for familiarity, p = 0.001 for recency;
vividness ratings: R2 = 0.733, F(2,45) = 61.844, p < 0.001; p < 0.001 for familiarity, p = 0.001
for recency).
3.3.3 Interview Details
The number of details in the descriptions of the scenes also differed significantly across the high
and low familiarity landmarks, as shown in Figure 4a. As with the ratings of detail, high
22
familiarity landmarks led to more detailed descriptions of scenes than did the low familiarity
landmarks (t(23) = 8.372, p < 0.001). Again, the number of details in the interviews correlated
significantly with the subjective ratings of detail, leading us to believe that subjects‟ self-ratings
were fairly accurate (r = 0.525, p < 0.001). Finally, a regression analysis using the number of
details from the interviews as the dependent factor revealed that familiarity was a significant
predictor of level of described detail, while recency was not (R2 = 0.381, F(2,45) = 13.856, p <
0.001; p = 0.001 for familiarity, p = 0.546 for recency).
For the interviews in the scene construction condition, it was also possible to assess the accuracy
of the details described, by comparing subjects‟ descriptions with images of the landmark in
question. Overall accuracy was found to be very high, with an average of 92% accurate details
for high familiarity scenes, and 85% for low familiarity scenes. The difference between the
accuracy scores approached significance (t(21) = 2.019, p = 0.056). If only the accurate details
were considered for the scene construction interviews, all findings remained the same: the
number of accurate details was significantly higher for scenes based on high familiarity
landmarks versus low familiarity landmarks (t(21) = 5.580, p < 0.001), the number of accurate
details correlated significantly with the participants‟ subjective ratings of detail (r = 0.465, p =
0.001), and a regression analysis revealed that familiarity, but not recency, was a significant
predictor of accurate details described (R2 = 0.327, F(2,41) = 9.958, p < 0.001; p = 0.002 for
familiarity, p = 0.867 for recency). Note that all of the above results are consistent with an alpha
level of 0.007 per comparison, in order to maintain a family-wise error rate below 0.05.
3.4 Imagination of the Future Condition
3.4.1 Retrieval Time
As in the other two conditions, it took less time to produce an imaginary experience if that event
was based around a highly familiar landmark versus a less familiar landmark. This difference
was confirmed by a paired t-test (t(25) = 3.241, p = 0.003), and is shown in Figure 2a. Similar to
the scene construction condition, this effect was consistent regardless of overall speed of
retrieval, and no median-split analyses were necessary.
23
3.4.2 Detail and Vividness Ratings
Following the same trends as the memory and scene conditions, imagined experiences were rated
as more detailed and more vivid if they were based on a highly familiar landmark, and less so if
they were based on a less familiar landmark (see Figure 3a and b). These differences were
verified with paired t-tests (t(25) = 5.528, p < 0.001 for detail ratings, t(25) = 5.633, p < 0.001 for
vividness ratings). Interestingly, the ratings of how close or far in the future the events were
imagined to occur also differed across the high and low familiarity landmark cues. If the cue was
a more familiar landmark, participants tended to place the imaginary event in the nearer future,
whereas imagined events based on less familiar landmarks tended to be imagined further away in
time (t(25) = 3.141, p = 0.004). Events based on high familiarity landmarks were given a mean
timeline rating of 2.72, where „2‟ represents 1-6 months and „3‟ represents 6-12 months into the
future, and events based on low familiarity landmarks were given a mean rating of 3.11 („4‟
represents more than one year but less than five years in the future).
For imaginary events, a measure of similarity to past memories was also collected. Imagined
events based on high familiarity landmarks tended to be slightly more similar to past memories
than events based on less familiar landmarks (t(25) = 2.639, p = 0.014), but this trend does not
reach significance if alpha levels are set at a more conservative level of 0.006 to control for
multiple t-tests and maintain the family-wise error rate at 0.05.
Regression analyses were performed to determine the predictive power of the familiarity of the
landmark, the proximity in the future of the event, and the similarity to past memories on the
ratings of detail and vividness of the imagined events. Both analyses yielded significant models,
showing familiarity as the most significant factor in predicting both the level of detail and
vividness of the imaginary event, whereas future proximity neared significance as a predictor,
and similarity to past memories was not a significant predictor (detail ratings: R2 = 0.344,
F(3,48) = 8.380, p < 0.001; p = 0.001 for familiarity, p = 0.055 for future proximity, p = 0.912
for similarity to past memories; vividness ratings: R2 = 0.324, F(3,48) = 7.665, p < 0.001; p =
0.002 for familiarity, p = 0.083 for future proximity, p = .720 for similarity to past memories).
24
3.4.3 Interview Details
As shown in Figure 4a, there was a trend in the interviews to describe imagined events based on
highly familiar landmarks in slightly more detail than those based on less familiar landmarks. A
paired t-test revealed that this trend did not reach significance at Bonferonni-corrected alpha
levels of 0.006 (t(25) = 2.562, p = 0.017). In addition, the correlation between interview details
and subjective ratings of detail was weak (r = 0.275, p = 0.048). However, closer inspection of
the data revealed differing trends depending on the amount of detail reported. As illustrated by
Figure 4b, a median-split of the sample showed that those who tended not to be very descriptive
(below the median number of details) did not show a difference in the amount of details
described for events based on high versus low familiarity landmarks (t(12) = 0.782, p = 0.449). In
contrast, those who were more descriptive in the interviews (above the median number of
details) showed a significant trend of more detail in descriptions of imaginary events based on
more familiar landmarks (t(12) = 4.372, p = 0.001). In addition, the highly-descriptive group
showed a strong correlation between subjective ratings of detail for imagined events and the
number of details described in the interviews about these events (r = 0.504, p = 0.009), whereas
the less-descriptive group showed no such correlation (r = 0.086, p = 0.677). These results may
indicate that a portion of participants were reluctant to provide full descriptions of their
imaginary events, and therefore showed no difference in described details across high and low
familiarity imagined events since the numbers of details were very low for both conditions.
However, the subjective ratings of detail still differed for these participants, leading us to believe
that the difference in amount of detail was still present, just not described, in these cases.
Finally, while a regression analysis based on the complete data set failed to produce a significant
model (R2 = 0.094, F(3,48) = 1.663, p = 0.188), the same analysis based on the highly-
descriptive half of the sample was significant, despite the reduction in power due to the smaller
group size. This regression revealed that, much like the analyses of subjective ratings of detail,
the familiarity of the landmark was a significant predictor of the amount of detail in the
interviews, the proximity in the future of the imaginary event approached significance as a
predictor, and the similarity to past memories was not a significant predictor (R2 = 0.338, F(3,22)
= 3.748, p = 0.026; p = 0.034 for familiarity, p = 0.079 for future proximity, p = 0.699 for
similarity to past memories).
25
As with the memory condition, a comparison of event-based versus scene-based details revealed
that, as instructed, participants successfully imagined and described event-based information and
reported very little scene information in this condition. This ensures that the trends observed in
each condition are unique to the task performed, not just replications of the same phenomenon.
3.5 Cross-Condition Comparisons
As is evident in Figures 2a, 3a and b, and 4a, when considering all participants in the experiment,
consistent trends were shown across all three tasks performed. Overall, participants took less
time to retrieve memories, scenes and imagined events based on highly familiar landmarks as
compared to low familiarity landmarks. Memories, scenes and imaginary events were
consistently more detailed and vivid if they were based on more familiar landmarks, and this was
true across both subjective and objective measures. Thus, in these respects, the effect of
familiarity was consistent across the three conditions in the study.
Additionally, when only the subjects who successfully completed two of the conditions in the
study were examined, it was found that performance was correlated across the different
conditions in the study. For the twenty subjects who completed both the imagination condition
and the memory condition, mean retrieval time in the memory condition was significantly
correlated with mean retrieval time in the imagination condition (r = 0.692, p < 0.001). This was
also true of detail and vividness ratings across the memory and imagination condition (r = 0.603,
p < 0.001 for detail ratings; r = 0.533, p < 0.001 for vividness ratings), and the number of details
described in the interviews across the two conditions (r = 0.582, p < 0.001). Similarly, for the
seventeen subjects who successfully completed the scene condition and the memory condition,
performance on these three measures was again correlated across conditions. The mean retrieval
time in the memory condition was significantly correlated with the mean retrieval time in the
scene condition (r = 0.540, p = 0.001), and this was also the case for detail ratings (r = 0.676, p <
0.001), vividness ratings (r = 0.678, p < 0.001), and the number of details described in the
interviews (r = 0.580, p < 0.001) across the memory and scene construction tasks.
Notably, although performance was correlated across conditions, and the direction of the effect
of familiarity was consistent across all three conditions, the size of this effect differed depending
on the task. Figure 5 illustrates that the magnitude of the difference across high and low
familiarity cues varied with task across measures of detail ratings, vividness ratings and
26
interview details. One-way ANOVAs confirmed this effect; task had a significant effect on the
magnitude of the difference between high and low familiarity cues in terms of detail ratings,
vividness ratings, and interview details (for detail ratings, effect of task: F(2,82) = 9.706, p <
0.001; for vividness ratings, effect of task: F(2,82) = 12.997, p < 0.001; for interview details,
effect of task: F(2,82) = 7.154, p = 0.001). There was no effect of task on the magnitude of the
retrieval time difference across high and low familiarity cues (F(2,82) = 1.244, p = 0.294).
As is evident from Figure 5a, b and c, the scene construction condition yielded larger differences
based on familiarity of the cue across all three measures (detail ratings, vividness ratings and
interview details). Post-hoc comparisons using Tukey‟s HSD test verified this effect, finding the
differences between detail ratings for high versus low familiarity landmarks significantly larger
in the scene construction condition than the memory condition (mean difference = 0.8115, p <
0.001) and the imagination condition (mean difference = 0.8394, p = 0.001), whereas the
differences in the memory and imagination condition did not differ from one another (mean
difference = 0.0279, p = 0.989). The same was true for vividness ratings (mean difference
between scene and memory = 0.9752, p < 0.001, mean difference between scene and imagination
= 0.8221, p < 0.001, mean difference between memory and imagination = 0.1531, p = 0.771) and
for the number of interview details (mean difference between scene and memory = 1.3909, p =
0.013, mean difference between scene and imagination = 1.8643, p = 0.001, mean difference
between memory and imagination = 0.4734, p = 0.316). In summary, while the familiarity of the
cue had the same effect on measures of detail and vividness across episodic memory, scene
construction, and imagination of the future tasks, this effect was consistently the strongest in the
scene construction condition.
4 Discussion
One goal of the present study was to determine the effect of the familiarity of a cue on the
retrieval time, detail-richness and vividness of memories, scenes and imagined events. We found
that across all three conditions in the experiment, more familiar cues led to faster retrieval times
for the remembered events, scenes and imaginary events. This finding was expected for scenes,
since we supposed that being more familiar with a location would lead to a richer, more
accessible representation of the scene around it, much like the frequency effects observed in
studies of word recognition (Scarborough et al., 1977). Since a scene is generally fairly static and
27
unchanging over time, there was no reason to believe that multiple encounters with a certain
scene would interfere with one another in any way; in fact, they should only serve to strengthen
the mental representation of that scene, and thus, make it more easily accessible, as we observed.
Following the predictions for the scene condition, we also expected that more experience with a
cue would lead to faster construction of possible future events since not only would the scene of
the event be more easily accessible but the participant would also have more memories
associated with that location which could be drawn upon in order to imagine a new event. This
prediction combined ideas from Hassabis and Maguire‟s (2007) scene construction theory with
the ideas from the constructive episodic simulation hypothesis (Addis & Schacter, 2008) by
theorizing that imagined events are constructed based on representations of scenes as well as
novel combinations of episodic details from memory. The results of this study supported this
view, since events were indeed imagined more quickly if they were based on more familiar
landmarks.
Our predictions for the memory condition were not as clearly defined. On the one hand, it was
possible that more familiar cues would mean more accessible memories, following the patterns
of the other two conditions. On the other hand, it was also possible that an increased number of
memories for a single cue would interfere with one another, creating a kind of „fan effect‟ and
therefore slower retrieval times (Anderson & Reder, 1999). Overall, results indicated that
multiple memories for a given cue did not interfere, and that isolating a single memory took less
time if one was more familiar with the landmark that served as the cue, following the same
pattern as the other two conditions. This finding may relate to pattern separation theories of
memory, which state that the hippocampus houses sparse, non-overlapping (or pattern-separated)
representations of memories (O‟Reilly & Norman, 2002). According to this theory, it is not
surprising that even the somewhat similar memories associated with a common cue do not
interfere with one another at retrieval, as we observed in this study. Perhaps, in this case, being
more familiar with a cue serves to make the whole set of related memories more readily
accessible, but due to their pattern-separated representations, does not entail any interference
between them.
It must be noted that some participants did not show faster retrieval times for memories based on
more familiar landmarks, and in particular, the few participants who took the longest to retrieve
28
memories showed a consistent reverse trend. In fact, every one of the five slowest participants
showed the opposite effect as compared with the majority of the participants in the study. This
led us to speculate that in the majority of cases, participants likely viewed the cue, quickly tried
to think of a memory associated with it, and pressed the button once they had one in mind, as
instructed. If the cue was more familiar, a memory would be more accessible and come to mind
faster, whereas if it was less familiar, it would take a little bit more time to bring a memory to
mind. The slowest participants, however, were perhaps employing a different search method,
causing more familiar cues to lead to slower reaction times, thus creating a fan effect. When the
cue appeared, it is possible that these participants attempted to call to mind all the memories that
were associated with that cue, and then once they had done so, selected one from the group, and
then pressed the button. This strategy would explain not only why they took longer in general,
since they were attempting to retrieve a larger number of memories instead of just a single one,
but also why more familiar cues would actually slow down retrieval instead of facilitating it. Of
course these are speculations based on the pattern observed in a small number of participants.
Further research is needed to determine if these differing search strategies are in fact employed,
and if they can explain the differing effects of familiarity observed in this study. However, it is
important to note that regardless of the retrieval time, the memories that were produced did not
differ from one another in terms of detail, vividness or recency of occurrence, which supports the
notion that what differed was retrieval strategy, and not factors relating to the content of the
memories.
We also observed the effects of cue familiarity on the quality of the memories, scenes and
imagined events in the study. We found robust differences in the detail-richness and vividness
across all three tasks, with higher familiarity cues leading to consistently more detailed and more
vivid scenes and events, whether remembered or imagined. This was true across subjective
ratings of detail and vividness as well as objective measures based on verbal descriptions, and in
the case of scenes, even when only accurate details were considered. The presence of this same
pattern across multiple measures of the quality of the memories, scenes and events leads us to
conclude that the familiarity of a cue has a very significant effect on the phenomenological
qualities of mental experience based on that cue. It is also significant that the scene construction
condition in the present study allowed for an assessment of accuracy of reported details. In most
studies involving memory for information originally acquired outside of the laboratory, it is very
29
difficult or even impossible to determine the veracity of reported memories. This condition
allowed us to determine not only that accuracy was very high for scene memories, but also that
all the observed effects were still shown when only accurate details were considered. This gives
us more confidence in the accuracy of subjects‟ memories overall and in the validity of the
findings from the present study and others on the topic of remote memory.
The effects of familiarity on the detail-richness and vividness of memories, scenes and imagined
future events were additionally found to exist independently of any effect of the recency of the
memory, the recency of the last visit to the scene, or the proximity or remoteness of the imagined
event. Previous studies have reported that more recent memories and events imagined closer in
time in the future are more detailed and are pre- or re-experienced more vividly (D‟Argembeau
& Van der Linden, 2004) and, similarly, that events based on more recently experienced contexts
also lead to more detailed and vivid imagined events (Szpunar & McDermott, 2008). This study
is the first to demonstrate that the cumulative experience one has with a cue is as important a
factor, if not a more important one, than recency in terms of the detail-richness and vividness of
memories, scenes, and imaginary events. Though recency was still shown to be a significant
factor in some cases, the overall familiarity of the participant with a cue exerted a greater effect
across all three tasks, and should be taken into consideration in future studies using cued memory
paradigms.
These findings are in line with the Construal Level Theory (Liberman & Trope, 2008), which
predicts that the level of detail of a mental representation follows from its “psychological
distance”. Objects or events that are „farther away‟ in terms of temporal, spatial or social
distance are thought of in less detailed terms, and „closer‟ objects and events tend to be
conceived of in more detail. The results of this study suggest that the familiarity of a cue is
another dimension of psychological distance, where more familiar landmarks led to more
detailed representations, and less familiar landmarks led to less detailed representations,
regardless of whether these representations were memories, scenes or imagined future events.
Another interesting finding was that imagined future events based on less familiar cues tended to
be placed further away in the future than imagined events based on more familiar cues. This
result also fits with the ideas from Construal Level Theory, which states that different
dimensions of psychological distance should be associated with one another. In this case, the
events that are „farther away‟ in terms of familiarity of the cue were placed farther away in time
30
as well. This same relationship was shown in the opposite direction by Arnold, McDermott and
Szpunar (2011), who demonstrated that imagined events in the near future were more likely to be
set in familiar locations than events in the far future. In their paradigm, participants were told
the timeline of the event but free to choose the location, whereas in the present study, the
location was the cue but the timeline was chosen freely.
Thus, theoretically, these results coincide well with idea that familiarity is another dimension of
psychological distance, according to Liberman & Trope‟s (2008) framework. They propose that
the mechanism behind these phenomena is such that events that are more „distant‟ on any
dimension are deliberately construed at a more abstract level since the more abstract details tend
to be less variable but specific details can only be known if the event is closer either in terms of
time, space, social distance, etc. In the present study, this may be true, particularly for future
events, since events based on high familiarity cues might be easier to imagine in more detail if
they occur somewhere the participant visits often or has planned to visit again soon. A related
explanation is that more experience with a cue means that there are more details associated with
that cue already in memory, and thus imagining or recollecting an event or scene based on that
cue can be done in more detail since there are more stored pieces of information upon which to
draw. This reasoning follows from Addis and Schacter‟s (2008) constructive episodic simulation
hypothesis, which postulates that memory and imagination are both constructive processes,
drawing on stored information to either reconstruct a previous memory or blend pieces of
information to create a novel event. Taken together, these two theories propose complementary
frameworks which not only explain the increased amount of detail and ratings of vividness for
memories, scene and imagined events based on more familiar cues, but also why the events tend
to be closer in time to the present than events based on less familiar landmarks.
The second goal of this study was to provide further insight into the similarities across episodic
memory, imagination of the future, and scene construction, and the relationships between them.
Despite the differing memory, scene, and imagination tasks performed in each condition, we
observed a similar effect of familiarity exhibited across all three, suggesting some commonality
or link between the tasks. These consistent trends were shown across measures of retrieval time,
as well as subjective and objective measures of detail and vividness, lending further support to
the notion that these three tasks are related, perhaps via a common underlying mechanism.
Furthermore, when participants successfully completed more than one condition in the study,
31
performance was correlated across the conditions, and this was again true of all measures,
including retrieval time, subjective detail and vividness ratings, and details described in the
interviews. The fact that the overall patterns, as well as individual performances, were consistent
across the varying tasks in the study buttresses the argument for shared underlying mechanisms
further still.
Of particular interest is the finding that the differences in detail and vividness ratings in high
compared to low familiarity trials were more pronounced in the scene construction condition.
This could be an indication that the representation of the scene is what varies the most as a result
of varying cue familiarity, and that the phenomenological differences observed in the episodic
memory and imagination of the future conditions stem from the scene representations also
involved in these more complex tasks. If the scene is the framework on which a memory or
imagined event is built, and a less familiar scene is less detailed and less vivid, this may explain
why the memories and imagined events based on less familiar cues also have fewer details and
are experienced less vividly. However, although we hypothesize that the memories and
imaginary future events are based on scenes, examination of the details reported in the interview
portion of the study confirmed that they possessed a richer event-based structure, and that
participants were in fact reporting very few scene-related details in their descriptions. This is
consistent with the notion that the process of scene construction underlies both episodic memory
and imagination of the future, but that these other abilities are more complex and also involve
additional processes. These ideas have been articulated in the scene construction hypothesis
advanced by Hassabis and Maguire (2007), and are supported by a variety of research (Hassabis,
Kumaran, Vann, & Maguire, 2007; Hassabis, Kumaran, & Maguire, 2007; Arnold, McDermott,
& Szpunar, 2011; Raffard, D‟Argembeau, Bayard, Boulenger, & Van der Linden, 2010;
Szpunar, Chan, & McDermott, 2009).
Ideas that the hippocampus is related to spatial representations have been present for decades. In
1978, O‟Keefe and Nadel proposed that the hippocampus functioned as a cognitive mapping
system both in animals and in humans. Since then, research has indicated that the hippocampus
may not be crucial for the retention of spatial maps, but is needed for the acquisition of such
maps, as well as detailed memories involving specific landmarks (Rosenbaum et al., 2000;
Rosenbaum et al., 2005; Hirshhorn et al., 2010; Hirshhorn, Grady, Rosenbaum, Winocur, &
Moscovitch, in press). Consistent with this is the idea that the hippocampus has evolved from a
32
purely spatial system, as may still be the case in certain animals, to a system for episodic and
autobiographical memories where representations of spatial context still play a crucial role
(Burgess, Maguire, Spiers, & O‟Keefe, 2001). This viewpoint complements the theory of scene
construction by providing an evolutionary context for it, as well as a way to reconcile it with
animal research.
While the present study provides evidence for a link between episodic memory, imagination of
the future, and scene construction, and the larger differences in the scene construction condition
support the notion of it as the underlying process, a limitation is that it demonstrates relation but
not overt reliance of the other two processes and that of scene construction. Another
interpretation of the results of this study could be that the tasks are related, but by virtue of all
tapping into some other underlying process common to all three. One possibility for this other
mechanism could be that the three tasks in the study rely on more general binding processes, not
scene construction specifically, to combine the spatial and/or episodic details into coherent
mental representations, as suggested previously by Rosenbaum et al. (2009). If this were the
case, the same results would be expected to be found as in this study, possibly with larger
differences in the scene construction condition due to the presence of more relationships in a
spatial description versus an episodic one. As shown in studies by Hoscheidt et al. (2010) and
Ryan et al. (2010), the hippocampus is preferentially involved in both episodic and spatial
processing, so perhaps studies that reveal a special relationship between the hippocampus and
scenes do so due to the presence of both episodic and spatial information involved in scenes, not
because scenes are the underlying mechanism involved in memory and imagination. Since it can
be difficult to inhibit the recollection of episodes when picturing scenes, and is nearly impossible
to remember episodes without any spatial context, it is a challenge for future research to attempt
to disentangle these two processes and determine if one is reliant on the other, or if both share
common underlying processes.
An additional direction for future research would be to compare the three tasks from the present
study – episodic memory, scene construction and imagination of the future – in a functional
neuroimaging paradigm. Significant overlap has been found between the areas involved in
episodic memory and imagination of the future, and these areas have been identified as part of
the default mode network (Addis et al., 2007; Spreng & Grady, 2010; Spreng et al, 2009 for
review). In addition, the network engaged during scene construction tasks also shares similar
33
areas, particularly the hippocampus, parahippocampal gyrus, retrosplenial and posterior parietal
cortices and the ventromedial prefrontal cortex (Hassabis et al., 2007). However, no study to date
has compared all three tasks in a highly matched paradigm such as the one used in the current
study. By controlling for other aspects of the tasks and comparing these three abilities directly,
finer observations of overlap or differentiation between the networks involved in each would be
able to be detected. It would also be possible to observe how the effects of familiarity of a cue
impact the areas of neuronal activity. It would be interesting to determine if the representation of
less familiar scenes, visited only once or twice, engaged more areas in common with episodic
memory whereas highly familiar scenes, which are more schematic in nature, may rely more on
other areas.
In conclusion, the present study provides future directions for research into the similarities and
differences across episodic memory, scene construction and imagination of the future, as well as
providing novel insights into the relationships between these three abilities and their interactions
with the familiarity of cues. By comparing these three tasks in a highly matched paradigm, it was
possible to observe and compare the effects of cue familiarity across the three tasks. This was the
first study to report that more cumulative experience with a cue results in faster retrieval of
individual memories associated with that cue, and that this effect is shown across retrieval of
scenes and construction of novel events as well. In addition, it showed that memories, scenes and
imagined events based on more familiar cues were both richer in detail and experienced more
vividly. These consistent results across the three tasks suggest a common mechanism uniting the
three, and the fact that the effects of familiarity were strongest in the scene construction
condition lends support to the notion that the mental representation of scenes is likely related to
that mechanism.
34
References
Addis, D. R., Cheng, T., P Roberts, R., & Schacter, D. L. (2011). Hippocampal contributions to
the episodic simulation of specific and general future events. Hippocampus. In press.
Addis, D. R., Musicaro, R., Pan, L., & Schacter, D. L. (2010). Episodic simulation of past and
future events in older adults: Evidence from an experimental recombination task.
Psychology and Aging, 25(2), 369-76.
Addis, D. R., Pan, L., Vu, M. A., Laiser, N., & Schacter, D. L. (2009b). Constructive episodic
simulation of the future and the past: Distinct subsystems of a core brain network mediate
imagining and remembering. Neuropsychologia, 47(11), 2222-38.
Addis, D. R., Sacchetti, D. C., Ally, B. A., Budson, A. E., & Schacter, D. L. (2009a). Episodic
simulation of future events is impaired in mild Alzheimer‟s disease. Neuropsychologia,
47(12), 2660-71.
Addis, D. R., & Schacter, D. L. (2008). Constructive episodic simulation: Temporal distance and
detail of past and future events modulate hippocampal engagement. Hippocampus, 18(2),
227-37.
Addis, D. R., Wong, A. T., & Schacter, D. L. (2007). Remembering the past and imagining the
future: Common and distinct neural substrates during event construction and elaboration.
Neuropsychologia, 45(7), 1363-77.
Addis, D. R., Wong, A. T., & Schacter, D. L. (2008). Age-Related changes in the episodic
simulation of future events. Psychological Science: A Journal of the American
Psychological Society, 19(1), 33-41.
Andelman, F., Hoofien, D., Goldberg, I., Aizenstein, O., & Neufeld, M. Y. (2010). Bilateral
hippocampal lesion and a selective impairment of the ability for mental time travel.
Neurocase: Case Studies in Neuropsychology, Neuropsychiatry, and Behavioural
Neurology, 16(5), 426-35.
Anderson, J. R. (1974). Retrieval of propositional information from long-term memory.
Cognitive Psychology, 6(4), 451-474.
Anderson, J. R., & Reder, L. M. (1999). The fan effect: New results and new theories. Journal of
Experimental Psychology: General, 128(2), 186-197.
Arnold, K. M., McDermott, K. B., & Szpunar, K. K. (2011). Imagining the near and far future:
The role of location familiarity. Memory & Cognition. In press.
Bonnici, H. M., Kumaran, D., Chadwick, M. J., Weiskopf, N., Hassabis, D., & Maguire, E. A.
(2011). Decoding representations of scenes in the medial temporal lobes. Hippocampus.
In press.
35
Botzung, A., Denkova, E., & Manning, L. (2008). Experiencing past and future personal events:
Functional neuroimaging evidence on the neural bases of mental time travel. Brain and
Cognition, 66(2), 202-12.
Buckner, R. L., & Carroll, D. C. (2007). Self-Projection and the brain. Trends in Cognitive
Sciences, 11(2), 49-57.
Burgess, N., Maguire, E. A., Spiers, H. J., & O'Keefe, J. (2001). A temporoparietal and
prefrontal network for retrieving the spatial context of lifelike events. Neuroimage, 14(2),
439-53.
Cooper, J. M., Vargha-Khadem, F., Gadian, D. G., & Maguire, E. A. (2011). The effect of
hippocampal damage in children on recalling the past and imagining new experiences.
Neuropsychologia. In press.
D'Argembeau, A., Raffard, S., & Van der Linden, M. (2008a). Remembering the past and
imagining the future in schizophrenia. Journal of Abnormal Psychology, 117(1), 247-51.
D'Argembeau, A., & Van der Linden, M. (2004). Phenomenal characteristics associated with
projecting oneself back into the past and forward into the future: Influence of valence and
temporal distance. Consciousness and Cognition, 13(4), 844-58.
D'Argembeau, A., & Van der Linden, M. (2006). Individual differences in the phenomenology of
mental time travel: The effect of vivid visual imagery and emotion regulation strategies.
Consciousness and Cognition, 15(2), 342-50.
D'Argembeau, A., Xue, G., Lu, Z. L., Van der Linden, M., & Bechara, A. (2008b). Neural
correlates of envisioning emotional events in the near and far future. Neuroimage, 40(1),
398-407.
Gaesser, B., Sacchetti, D. C., Addis, D. R., & Schacter, D. L. (2011). Characterizing age-related
changes in remembering the past and imagining the future. Psychology and Aging, 26(1),
80-84.
Gamboz, N., De Vito, S., Brandimonte, M. A., Pappalardo, S., Galeone, F., Iavarone, A., &
Della Sala, S. (2010). Episodic future thinking in amnesic mild cognitive impairment.
Neuropsychologia, 48(7), 2091-7.
Hassabis, D., Kumaran, D., & Maguire, E. A. (2007). Using imagination to understand the neural
basis of episodic memory. The Journal of Neuroscience: The Official Journal of the
Society for Neuroscience, 27(52), 14365-74.
Hassabis, D., Kumaran, D., Vann, S. D., & Maguire, E. A. (2007). Patients with hippocampal
amnesia cannot imagine new experiences. Proceedings of the National Academy of
Sciences, 104(5), 1726.
36
Hassabis, D., & Maguire, E. A. (2007). Deconstructing episodic memory with construction.
Trends in Cognitive Sciences, 11(7), 299-306.
Hassabis, D., & Maguire, E. A. (2009). The construction system of the brain. Philosophical
Transactions of the Royal Society of London. Series B, Biological Sciences, 364(1521),
1263-71.
Hirshhorn, M., Grady, C.L., Rosenbaum, R.S., Winocur, G., & Moscovitch, M. (2010). Brain
regions involved in the retrieval of spatial and episodic details associated with a familiar
environment: an fMRI study. Submitted.
Hirshhorn, M., Grady, C., Rosenbaum, R., Winocur, G. and Moscovitch, M. (2011). The
hippocampus is involved in mental navigation for a recently learned, but not a highly
familiar environment: A longitudinal fMRI study. Hippocampus. In press.
Hirshhorn, M., Newman, L., & Moscovitch, M. (2011). Detailed descriptions of routes traveled,
but not map-like knowledge, correlates with tests of hippocampal function in older adults.
Hippocampus. In press.
Hoscheidt, S. M., Nadel, L., Payne, J., & Ryan, L. (2010). Hippocampal activation during
retrieval of spatial context from episodic and semantic memory. Behavioural Brain
Research, 212(2), 121-32.
Klein, S. B., Loftus, J., & Kihlstrom, J. F. (2002). Memory and temporal experience: The effects
of episodic memory loss on an amnesic patient's ability to remember the past and imagine
the future. Social Cognition, 20(5), 353-379.
Kwan, D., Carson, N., Addis, D. R., & Rosenbaum, R. S. (2010). Deficits in past remembering
extend to future imagining in a case of developmental amnesia. Neuropsychologia,
48(11), 3179-3186.
Levine, B., Svoboda, E., Hay, J. F., Winocur, G., & Moscovitch, M. (2002). Aging and
autobiographical memory: Dissociating episodic from semantic retrieval. Psychology and
Aging, 17(4), 677-89.
Liberman, N., & Trope, Y. (2008). The psychology of transcending the here and now. Science
(New York, N.Y.), 322(5905), 1201-5.
Maguire, E. A., Vargha-Khadem, F., & Hassabis, D. (2010). Imagining fictitious and future
experiences: Evidence from developmental amnesia. Neuropsychologia, 48(11), 3187-92.
Malmberg, K. J., Steyvers, M., Stephens, J. D., & Shiffrin, R. M. (2002). Feature frequency
effects in recognition memory. Memory & Cognition, 30(4), 607-13.
Monsell, S., Doyle, M. C., & Haggard, P. N. (1989). Effects of frequency on visual word
recognition tasks: Where are they? Journal of Experimental Psychology. General, 118(1),
43-71.
37
Morrison, C. M., & Ellis, A. W. (1995). Roles of word frequency and age of acquisition in word
naming and lexical decision. Journal of Experimental Psychology: Learning, Memory
and Cognition, 21(1), 116-133.
Nyberg, L., Kim, A. S., Habib, R., Levine, B., & Tulving, E. (2010). Consciousness of subjective
time in the brain. Proceedings of the National Academy of Sciences of the United States
of America, 107(51), 22356-9.
O'Keefe, J., & Nadel, L. (1978). The hippocampus as a cognitive map. Clarendon Press Oxford.
O'Reilly, R. C., & Norman, K. A. (2002). Hippocampal and neocortical contributions to
memory: Advances in the complementary learning systems framework. Trends in
Cognitive Sciences, 6(12), 505-510.
Okuda, J., Fujii, T., Ohtake, H., Tsukiura, T., Tanji, K., Suzuki, K., Kawashima, R., Fukuda, H.,
Itoh, M., & Yamadori, A. (2003). Thinking of the future and past: The roles of the frontal
pole and the medial temporal lobes. Neuroimage, 19(4), 1369-80.
Race, E., Keane, M. M., & Verfaellie, M. (2011). Medial temporal lobe damage causes deficits
in episodic memory and episodic future thinking not attributable to deficits in narrative
construction. The Journal of Neuroscience: The Official Journal of the Society for
Neuroscience, 31(28), 10262-9.
Raffard, S., D'Argembeau, A., Bayard, S., Boulenger, J. P., & Van der Linden, M. (2010). Scene
construction in schizophrenia. Neuropsychology, 24(5), 608-15.
Rosenbaum, R. S., Gao, F., Richards, B., Black, S. E., & Moscovitch, M. (2005). "Where to?"
Remote memory for spatial relations and landmark identity in former taxi drivers with
Alzheimer‟s disease and encephalitis. Journal of Cognitive Neuroscience, 17(3), 446-62.
Rosenbaum, R. S., Gilboa, A., Levine, B., Winocur, G., & Moscovitch, M. (2009). Amnesia as
an impairment of detail generation and binding: Evidence from personal, fictional, and
semantic narratives in K.C. Neuropsychologia, 47(11), 2181-7.
Rosenbaum, R. S., Priselac, S., Köhler, S., Black, S. E., Gao, F., Nadel, L., & Moscovitch, M.
(2000). Remote spatial memory in an amnesic person with extensive bilateral
hippocampal lesions. Nature Neuroscience, 3(10), 1044-8.
Rosenbaum, R. S., Ziegler, M., Winocur, G., Grady, C. L., & Moscovitch, M. (2004). "I have
often walked down this street before": Fmri studies on the hippocampus and other
structures during mental navigation of an old environment. Hippocampus, 14(7), 826-35.
Ryan, L., Lin, C. Y., Ketcham, K., & Nadel, L. (2010). The role of medial temporal lobe in
retrieving spatial and nonspatial relations from episodic and semantic memory.
Hippocampus, 20(1), 11-8.
38
Scarborough, D. L., Cortese, C., & Scarborough, H. S. (1977). Frequency and repetition effects
in lexical memory. Journal of Experimental Psychology: Human Perception and
Performance, 3(1), 1-17.
Schacter, D. L., & Addis, D. R. (2009). On the nature of medial temporal lobe contributions to
the constructive simulation of future events. Philosophical Transactions of the Royal
Society of London. Series B, Biological Sciences, 364(1521), 1245-53.
Schacter, D. L., Addis, D. R., & Buckner, R. L. (2007). Remembering the past to imagine the
future: The prospective brain. Nature Reviews. Neuroscience, 8(9), 657-61.
Spreng, R. N., & Grady, C. L. (2010). Patterns of brain activity supporting autobiographical
memory, prospection, and theory of mind, and their relationship to the default mode
network. Journal of Cognitive Neuroscience, 22(6), 1112-23.
Spreng, R. N., & Levine, B. (2006). The temporal distribution of past and future
autobiographical events across the lifespan. Memory & Cognition, 34(8), 1644-51.
Spreng, R. N., Mar, R. A., & Kim, A. S. (2009). The common neural basis of autobiographical
memory, prospection, navigation, theory of mind, and the default mode: A quantitative
meta-analysis. Journal of Cognitive Neuroscience, 21(3), 489-510.
Szpunar, K. K., Chan, J. C., & McDermott, K. B. (2009). Contextual processing in episodic
future thought. Cerebral Cortex, 19(7), 1539-48.
Szpunar, K. K., & McDermott, K. B. (2008). Episodic future thought and its relation to
remembering: Evidence from ratings of subjective experience. Consciousness and
Cognition, 17(1), 330-4.
Szpunar, K. K., Watson, J. M., & McDermott, K. B. (2007). Neural substrates of envisioning the
future. Proceedings of the National Academy of Sciences of the United States of America,
104(2), 642-7.
Tulving, E. (1985). Memory and consciousness. Canadian Psychology, 26, 1-12.
Weiler, J. A., Suchan, B., & Daum, I. (2010). When the future becomes the past: Differences in
brain activation patterns for episodic memory and episodic future thinking. Behavioural
Brain Research, 212(2), 196-203.
39
Figures
Figure 1.
Figure 1. Representation of one trial in the episodic memory condition of the experiment. Order
of slides and duration is indicated in the upper left corner of each frame.
1 – 2 secs. 2 – max. 10 secs.
3 – 20 secs. 4 – max. 30 secs.
5 – max. 30 secs. 6 – max. 30 secs.
40
Figure 2. Retrieval Time.
Figure 2a. Mean retrieval time (in seconds) for memories, scenes, and imagined events based on
high and low familiarity landmarks. Error bars indicate the standard error of the mean for each
group.
Figure 2b. Mean retrieval time (in seconds) for memories based on high and low familiarity
landmarks, in the group of participants below the median retrieval time, and the group above the
median retrieval time. Error bars indicate the standard error of the mean for each group.
41
Figure 3. Detail and Vividness Ratings.
Figure 3a. Mean ratings of detail (on a 1-5 scale) for memories, scenes, and imagined events
based on high and low familiarity landmarks. Error bars indicate the standard error of the mean
for each group.
Figure 3b. Mean ratings of vividness (on a 1-5 scale) for memories, scenes, and imagined events
based on high and low familiarity landmarks. Error bars indicate the standard error of the mean
for each group.
42
Figure 4. Number of Interview Details.
Figure 4a. Mean number of details described per memory, scene, and imagined event based on
high and low familiarity landmarks. Error bars indicate the standard error of the mean for each
group.
Figure 4b. Mean number of details described per imagined event based on high and low
familiarity landmarks, in the group of participants below the median number of details, and the
group above the median number of details. Error bars indicate the standard error of the mean for
each group.
43
Figure 5. Magnitude of differences across high and low familiarity landmarks.
Figure 5a. Mean magnitude of difference in detail ratings across high and low familiarity
landmarks, across memory, scene and imagination conditions. Error bars represent standard error
of the mean for each group.
Figure 5b. Mean magnitude of difference in vividness ratings across high and low familiarity
landmarks, across memory, scene and imagination conditions. Error bars represent standard error
of the mean for each group.
44
Figure 5c. Mean magnitude of difference in number of interview details described across high
and low familiarity landmarks, across memory, scene and imagination conditions. Error bars
represent standard error of the mean for each group.
45
Appendices
Appendix 1. Full list of Toronto landmarks used in study.
1 1 Spadina Circle (old Knox College)
2 Air Canada Centre
3 Allan Gardens Conservatory
4 Art Gallery of Ontario
5 Atrium-on-Bay
6 Banting Institute
7 Bata Shoe Museum
8 Bloor Cinema
9 Canada's National Ballet School
10 Canon Theatre (Pantages Theatre)
11 Casa Loma
12 CBC Broadcast Centre
13 Chinatown Centre
14 City Hall (Nathan Phillips Square)
15 City TV Building (MuchMusic)
16 Clarke Institute (CAMH)
17 CN Tower
18 College Park
19 Commerce Court
20 Convocation Hall
21 Dundas Square
22 Eaton Centre
23 Elgin and Winter Garden Theatre
24 U of T Exam Centre
25 First Canadian Place (Toronto Stock Exchange)
26 Flatiron Building (Gooderham Building)
27 Flavelle House (Law Library)
28 Four Seasons Centre for the Performing Arts
29 Four Seasons Hotel (Yorkville)
30 Gladstone Hotel
31 Graduate House
32 Greyhound Bus Terminal
33 Hart House
34 Health Sciences Building
35 Hilton Hotel (University and Richmond)
36 Hockey Hall of Fame (BCE Place)
37 Holt Renfrew on Bloor
38 Honest Ed‟s
39 Hospital for Sick Children
40 Hudson‟s Bay Company (Yonge and Queen)
41 Koffler Student Services Building
46
42 Lash Miller Chemical Laboratories
43 Lee's Palace
44 Leslie L. Dan Pharmacy Building
45 Manulife Centre (Bay and Bloor)
46 Maple Leaf Gardens
47 MaRS Centre
48 Massey College
49 Massey Hall
50 Medical Sciences Building
51 Metro Toronto Convention Centre
52 Mount Sinai Hospital
53 OCAD Building (Sharp Centre for Design)
54 OISE building
55 Old City Hall
56 Old Toronto Stock Exchange Building (Design Exchange)
57 Osgoode Hall
58 Planetarium (Children‟s Own Museum)
59 Princes' Gate
60 Princess Margaret Hospital
61 Princess of Wales Theatre
62 Queen‟s Park (Parliament Buildings)
63 Queen‟s Quay Terminal (Harbourfront Centre)
64 Redpath Sugar Museum
65 Ricoh Coliseum
66 Robart‟s Library
67 Rogers Centre (Skydome)
68 Roy Thompson Hall
69 Royal Alexandra Theatre
70 Royal Bank Plaza
71 Royal Conservatory of Music
72 Royal Ontario Museum
73 Royal York Hotel
74 Sandford Fleming Engineering Building
75 Scotia Plaza (King and Bay)
76 Scotiabank Theatre (Paramount)
77 Second City Theatre
78 Sheraton Centre
79 Sidney Smith Hall
80 Silvercity Yonge-Eglinton Theatre
81 Sony Centre for the Performing Arts (Hummingbird/O‟Keefe Centre)
82 St. George Subway Station (St George Street entrance)
83 St. James Cathedral
84 St. Lawrence Market
85 St. Michael‟s Hospital
86 St. Patrick‟s Church
87 Steamwhistle Brewery/Roundhouse Building
47
88 The Brunswick House
89 The Drake Hotel
90 The Madison Avenue Pub
91 Toronto Dominion Centre
92 Toronto General Hospital
93 Toronto Island Ferry Terminal
94 Toronto Police Museum and Discovery Centre
95 Trinity College
96 U of T Athletic Centre
97 Union Station
98 University College
99 U of T Student's Union Building (Louis B. Stewart Observatory)
100 Varsity Stadium
101 Victoria College
102 Westin Harbour Castle Hotel
103 Woodsworth College Residence
104 Miles Nadal Jewish Community Centre (Spadina and Bloor)
105 Christie Pitts Park
106 Riverdale Park
107 Metro Toronto Zoo
108 Ashbridges Beach
109 Ontario Place (entrance)
110 High Park
111 Ontario Science Centre (entrance)
112 Old Mill Inn
48
Appendix 2. Interview Coding Guide.
(Adapted from Levine‟s Autobiographical Interview scoring manual.)
In this study, subjects are asked to describe memories, visual scenes, and imaginary future events
based on certain Toronto landmarks with which they are familiar. The events should be specific
in time and place, and the scenes should refer only to visual, non-event-specific features of the
landmark and its surroundings. Some general probing is used at times to elicit full descriptions,
but no specific probes (i.e. questions related to specifics) are used.
These interviews are then transcribed, and the number of relevant details for each scene,
memory, or imaginary event is counted. We are only interested in details central to the event or
scene in question, not extraneous details, background information or comments on the memories
or scenes.
Memories
What counts as a detail:
- One bit of information regarding the memory being described: an occurrence,
observation, fact, statement or thought
- Consider whether that part of the text contributes information to the story being told
- The details must all be “internal”, that is, pertaining directly to the event being described
- We don‟t count “external” details, i.e. events not related to the main event being
described, semantic or factual information
- Be careful of sentences using „because‟ or „since‟:
o “{we went to McDonalds} because {my friend Tom was starving}” = 2 details
o “{we went to McDonalds} because we always go there” = 1 detail + 1 semantic
- Details include:
o actions or events that happened in the memory
o who was there ({I went there to see a movie} {with my mom} = 2 details, event
+ people; but {I went there with my mom} = 1 detail – no specific event)
o emotions, thoughts, feelings, expectations, beliefs felt or expressed at the time
o inferences about the mental states of others (made at the time of the memory)
o the weather
o the time (of day, or year)
o people‟s clothing
o the location of things or people in the memory (scene-type details count, even if
they might be a bit more „semantic‟ in nature)
o where the memory occurred, what was around
o any sensory perceptions (sounds, smells, tastes, feelings)
o anything seen during the event (objects, buildings, colours)
o quotes or conversations (each statement = one detail)
49
- Things that are not counted as details include:
o thoughts expressed in retrospect, reflections or judgements of the memory (“It
was pretty cool”; but “I thought it was cool (at the time)” counts)
o relative statements like “that was the first time I‟d been there”, “that was one of
the best times I‟d been there”, since that is comparing the memory to other
memories and not as much part of the memory itself
o long-standing beliefs or opinions (“I‟ve always loved cheese”)
o any general knowledge or facts
o any unrelated events or stories
o any repeated details (i.e. “we went there in the winter… it was the winter, so”;
only the first mention counts as a detail)
o metacognitive statements (i.e. “let‟s see what else I can remember”)
o inferences (i.e. “It must have been the morning since that‟s when I always go
there”)
Scenes
- Only visual/spatial information about the landmark itself, the surroundings, or the
appearance of the area is counted
- Details of the building, colours, locations of doors, windows, interior/exterior features,
etc. all count as details
- Nearby buildings or sights can also count as details
- Other aspects of a visual scene count as details even if they‟re not always a part of it, i.e.
hot-dog vendor on the corner, construction (these can count because they are fairly stable
elements and may be part of the mental representation of the scene)
- Event-specific details or objects should not be counted (last time I was there it was very
busy, it used to be under construction, my friend was standing there)
- Statements that don‟t count include:
o Event-related information
o Weather
o Facts about the place
o Reflections, thoughts, feelings associated with the place
o Any non-visual information
o Street names or general area (semantic, could be based on maps)
o Any descriptions that are not, or could not be, part of the scene (i.e. things that are
“nearby” but actually not visible from the landmark, or very vague statements like
“I know there‟s a McDonald‟s near there somewhere…”)
o Very general statements (“it‟s nice”)
50
Imaginary Events
- Imaginary events should be coded in the same way as memories
- Statements preceded by “I guess”, “it would/might be”, “I‟m going to”, “probably”, or
“hopefully” should be taken as „factual‟ for the purpose of the imaginary events
- Use the same rules for details as the memories
Examples: (Phrases between square brackets count as one detail, footnotes explain omissions.)
Memory: Robart‟s library.
“The [first time I had to do an essay here, at UofT], that was [last January I think]. [I‟d never
seen a library so big]. [So I went in, was in a complete state of confusion and disarray] and so
forth. I was [trying to match the letters, to find a book] but [I was on the wrong floor]. [I found
my way around it], I‟m good at it now1.”
1. General information, not related to remembered event
Scene: U of T Exam centre.
“I had my exam there, so it‟s on McCaul street1 which is not a major thoroughfare as well
1, so
when you enter, [if you‟re walking down from north to south it‟s on the left side of the street],
which means it‟s on the east side of the street2. When you enter the building, [there‟s a small
flight of stairs for you to go up.] And then the [waiting area is mostly empty], [there are benches
on the side, along the walls] and [there‟s one exam room on the right] and you go further in, and
[there‟s another exam room], and as you enter the exam room [there are the washrooms, the male
washroom first] and [then the female washroom into the exam room]. There are [upstairs, the
building actually has a couple of floors] but I‟ve only been to the first floor where I had my
exams in3.”
1. Semantic or map-based knowledge
2. Inference based on other information
3. Event-related information
Imagination: Union station.
“I imagined that [it was my friend‟s birthday in Montreal] and [I had to get there right after
work] and [it was completely packed] and [it was getting frustrating] [everyone was running here
and there]. I was getting frustrated1 [I couldn‟t find my train] and [after about half an hour of
running around, I finally found it] and [then I boarded my train].”
1. Repetition