University of Southern California Dissertations and Theses

Neural processing of temporal distance in film

(USC Thesis Other)

Neural processing of temporal distance in film

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content NEURAL PROCESSING OF TEMPORAL DISTANCE IN FILM
by
Chelsey Pan
A Thesis Presented to the
FACULTY OF THE USC DANA AND DAVID DORNSIFE
COLLEGE OF LETTERS, ARTS, AND SCIENCES
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(PSYCHOLOGY)
December 2024
Copyright 2024 Chelsey Pan

ii
TABLE OF CONTENTS
List of Figures…………………………………………………………………………………….iii
Abstract…………………………………………………………………………………………...iv
Chapter One: Introduction..…………………………...…………………………………………..1
Chapter Two: Hypotheses………………………….…………………………………………..….4
Chapter Three: Methods……..……………………………………………………………………6
Chapter Four: Results……………………………………………………………………………11
Chapter Five: Discussion…………………………………………….…………………………..16
Chapter Six: Limitations and Future Directions…………………...…………………………….18
References………………………………………………………………………………………..20

iii
List of Figures
Figure 1. Camera Cuts Per Time Shift…………………………………………………………...11
Figure 2. Hidden Markov Model Results: Visual and Auditory Regions………………………..12
Figure 3. Hidden Markov Model Results: Angular Gyrus………………………………………13
Figure 4. Hidden Markov Model Results: Medial Prefrontal Cortex (mPFC)…………………..13
Figure 5. Hidden Markov Model Results: Posteromedial Cortex (PMC)……………………….14
Figure 6. Permutation Test Results………………………………………………………………15

iv
Abstract
Films are rarely entirely linear and unbroken, often leaving viewers to fill in gaps and
temporally rearrange content in order to keep track of an ongoing narrative. The Default Mode
Network (DMN) has previously been implicated in mental time travel and the formation of event
schemas, however it remains unclear how the DMN responds to variations in temporal distance
during narrative time shifts. Drawing from Event Segmentation Theory and current models of
predictive processing involved in narrative comprehension, we used Hidden Markov Models to
examine brain activity during event boundaries for regions of the brain located within the DMN
and lower level sensory cortical regions while watching Forrest Gump. Our results demonstrated
early evidence that while areas in the early visual and early auditory cortices responded
indiscriminately to all changes in information regardless of narrative significance, brain areas
located in the DMN tended to respond more reliably to longer narrative time shifts. In addition to
providing further evidence for a hierarchy of temporal processing in the brain, these findings
demonstrate that the DMN may be involved in stitching together temporally discontinuous
events in order to form a continuous narrative.

1
Chapter One: Introduction
Stories are nearly always told in parts. With relatively few exceptions, a television show
is broken up into seasons and episodes, a book charts a narrative arc over many chapters, and a
film has its classic three act structure (Field, 2005). Each of these parts can typically be broken
down even further, such as a single film act into individual scenes and shots. However, films are
rarely completely linear and unbroken; interstitial content between shots often excludes
considerable amounts of information and can vary dramatically in terms of how much time
within the narrative has passed. Two shots that cut between characters speaking may have little
to no time passing between them at all, while another edit that cuts between a character’s
childhood and adulthood can have decades’ worth of time shifting past.
These distinct and often discontinuous chunks can also be thought of as “events,” which
are segments of time that are seen as having a defined beginning and end (Zacks, 2001).
According to Event Segmentation Theory (EST), the brain naturally engages in this process of
event segmentation during information processing, where event boundaries are created due
situational or motivational changes in the context or characters during a narrative (Zacks et al.,
2007).
The extent we attend to these events, and the level of abstraction at which we consider
them, can vary depending on various individual differences in interpretation and organizing
features (Sava-Segal et al., 2023). This event processing thus occurs across multiple timescales;
an event can vary in duration so long as the five dimensions of the event - the current time, the
current location (space), the objects, the characters, the causes, and the intentions of the
protagonists - persist (Zacks et al., 2007). Neural evidence for event representation in narratives
shows that events are also processed on multiple timescales within the brain (Baldassano et al,

2
2017), with longer timescales corresponding to higher order brain regions that are associated
with more complex cognition (Lerner et al., 2011).
However, it remains unclear how the brain responds to and makes sense of narrative time
shifts while watching films. One possibility is that keeping track of time shifts involves the
integration of both prior knowledge and ongoing prediction in order to continuously make sense
of a shifting timeline. EST posits that event segmentation in narratives involves the construction
of event schemas, wherein audiences use prior knowledge in order to segment narratives into
discrete events (Zacks et al., 2007). Audiences overlay these situational models onto the
narrative in order to make predictions about events and event boundaries. Neuroimaging studies
employing the use of Hidden Markov Models have found evidence for the representation of
event schemas while processing short narratives in both audiovisual and purely auditory format
(Baldassano et al., 2018).
The brain’s Default Mode Network (DMN) has been implicated in the representation of
acquired schemas (Gilboa & Marlatte, 2017); it has specifically been found to represent schemas
during both narrative processing and recall (Lee, Bellana, & Chen, 2020). The DMN is a system
of regions in the brain (angular gyrus, medial prefrontal cortex, posteromedial cortex) that seems
to be involved in integrating intrinsic and extrinsic information over long time scales in order to
make sense of continually unfolding events (Yeshurun, Nguyen, & Hasson, 2021).
More broadly, the DMN has been shown to be involved in a wide range of processes,
including mental time travel, with functional connectivity in the DMN associated with both
reminiscing about the past and imagining about the future (Schacter & Addis, 2007; Østby et al.,
2012). Recent work has highlighted the DMN’s role in predictive processing, positing mental
time travel as a unidirectional process of event simulation where past events are reconstructed

3
rather than genuinely recalled; this mechanism ultimately facilitates predictive thinking (Addis,
2020). Regions such as the DMN, which are higher up on the cognitive processing hierarchy and
thus process information over longer timescales, have been suggested to engage in predictive
coding, where information is considered in light of prior context and used to make predictions
about future events (Lerner et al., 2011).
While the DMN is involved in both mental time travel and schematic representations of
events, it’s uncertain how the DMN and other regions of the brain respond to shifts in time
between narrative event transitions within a film. Additionally, little is known about how the
DMN responds to variations in temporal distance during narrative time travel. The current
research aimed to elucidate the role of the DMN in integrating disparate events across time to
form a continuous narrative.

4
Chapter Two: Hypotheses
We hypothesized that higher order areas of the brain associated with meaning
representation and mental time travel would be more sensitive to temporal transitions that
significantly change the meaning or content of the narrative, rather than solely visual or locationbased transitions. Because the DMN is involved in meaning-making, it is thought to be
insensitive to low-level perceptual changes that don’t involve changes in meaning (Yeshurun,
Nguyen, & Hasson, 2021). We predicted, then, that it would similarly be insensitive to camera
shifts that change what the viewer literally observes, but makes no meaningful transitions in the
narrative or characters’ mental states.
Additionally, as larger amounts of time shift within a narrative, the content of the story
changes more and more significantly. Perceptually, characters change in age and appearance,
while physical locations are rearranged or altered entirely. More critically still, the goals and
motivations of the characters and their relationship to each other and the broader narrative shift
dramatically. As such, we expect that greater changes in the meaning of the story occur during
these temporal shifts, which higher order systems of the brain would attend more strongly to.
Larger transitions in time within the narrative are thus predicted to lead to larger transitions
between brain states in higher order areas such as the DMN, as it works harder to reconcile the
disparate states.
It is possible that as time jumps between cuts grow, it becomes more difficult to make
ongoing predictions about narrative events. There is some evidence that keeping track of time is
cognitively taxing - prior research has found that reading sentences that contained a shift in
narrative time by one hour were more difficult to read and more difficult to recall than reading
sentences where no shift in narrative time occurred (Speer & Zacks, 2005). Comparatively, we

5
expect lower areas of the brain such as the visual and auditory cortices, which are not thought to
be involved in this predictive or meaning-based processing, to respond indiscriminately to all
visual and auditory fluctuations, regardless of whether they involve a temporal transition.
In this study, we examined neuroimaging data from people watching the film Forrest
Gump. We used Hidden Markov Models (HMM) to identify boundaries between stable neural
states, and compared the best fitting number of event boundaries identified across different brain
regions. We focused specifically on comparing the number of event boundaries established in
regions of the DMN to event boundaries in the early visual and auditory cortical regions. Based
on our initial hypotheses, we expected that the HMM would fit better with a smaller number of
event boundaries in the DMN and a greater number of boundaries in the visual and auditory
cortices.

6
Chapter Three: Methods
Participants: Analyses for this study were conducted on a portion of the StudyForrest
dataset (Hanke et al., 2016), an open source project focused on collecting neuroimaging data
while participants experience the film Forrest Gump. 15 participants in Germany watched a
slightly shortened version of Forrest Gump inside the scanner.
Stimulus: Participants watched Forrest Gump (1994), a film that follows the eponymous
Forrest through multiple decades of his life as he finds himself inadvertently playing a major role
in shaping myriad key moments in United States history. Forrest Gump was a particularly
suitable choice for this research question due to its extensive number of unmarked narrative time
shifts, which varied in both length and direction across the entire duration of the film.
The audio for the film was dubbed into German and shown to participants across 8
roughly 15 minute long segments. Between segments, there were several seconds of fade in and
out onto a gray screen. After 4 segments, participants exited the scanner for an untimed break
before returning to watch the final 4 segments. On average, participant breaks lasted around 15
minutes.
Neuroimaging data: Although the participants watched the film in 8 separate chunks, all
following analyses were conducted on the full combined film scans. This was primarily done in
order to maximize the number of cuts of different time spans in the analysis. For each
participant, the last 8 volumes of their functional imaging scans for the first 7 runs were cut in
order to remove portions of the video that had overlapping or fading out scenes and then the 8
runs were stitched together. Each TR was 2000ms long, and the total length of the scan after
stitching the 8 segments together was 3543 volumes. Data manipulation was conducted in
Python 2 using the PyMVPA package (Haxby et al., 2011).

7
Annotations: Four (3 primary, 1 pilot) coders within the Brain and Creativity Institute
watched each of the 8 Forrest Gump segments in order and annotated the timestamps for every
cut within each segment. Although the segments were identical in timing to those shown to
participants in Germany, the Coders watched the segments with the original English audio.
For each cut, participants also estimated how much time within the narrative had passed
between cuts, using a scale ranging from 0 (meaning that no time has passed) to 7 (meaning
several years have elapsed). Table 1 shows a table of the scale coders used to categorize each cut
and its corresponding time shift. Whenever applicable, coders also indicated the direction of the
time shift as being either forward or backward. Cuts were defined as any instance of
discontinuity in the camera shot.
Time Shift Code Time Elapsed Between Cuts
0 No time passed
1 Seconds
2 Minutes
3 Hours
4 Days
5 Weeks
6 Months
7 Years
Table 1. A table of time shift codes used to annotate all of the cuts in the stimuli. The right column displays the
corresponding amount of temporal distance passing between cuts. As an example, all cuts marked with a “7” are
cuts where more than one year passed between the shots within the narrative.

8
Values for timestamps, time shifts, and direction were determined based on agreement
from at least 2 out of the 3 primary coders. In the event that a coder listed a timestamp that others
did not, this was double checked against the additional pilot coder. Values with no agreement
among the coders were individually rechecked against the film to make a final determination.
Analyses: A series of Hidden Markov Models (HMM) were run on six regions of interest
(ROI). The six ROIs included the left and right angular gyrus, the medial prefrontal cortex
(mPFC), the posteromedial cortex (PMC), and the early visual and early auditory cortices. The
mPFC, PMC, and angular gyrus were chosen as prominent regions of the DMN (Raichle, 2015),
while the visual and auditory regions served as lower level sensory comparisons. To account for
variations in the size of these brain areas, 8 millimeter spheres each containing 257 voxels were
selected from each region to form all of our ROIs. The location of the spheres was chosen from a
previous analysis of resting state data from 76 participants in our laboratory. We chose the peak
coordinates from angular gyrus, posterior medial cortex, and medial prefrontal cortex from this
analysis. Visual and auditory cortex coordinates were chosen based on anatomy: the visual
cortex sphere was centered along the calcarine sulcus at the midline, and the auditory cortex
sphere was centered on left Heschl’s gyrus.
The HMM was used to determine which among a set of predetermined number of total
possible event boundaries (K values) best fit the data by identifying changes in stable neural
patterns, signaling a transition from one event to another. The HMMs tested a series of supplied
K values to determine the best fitting number of event boundaries for each ROI, which were
obtained from the finalized Forrest Gump film annotations. The Ks were selected based on the
number of cuts in each scale of time shift. For example, the first K was the number of cuts that
had been rated as 7 (several years had passed), then the second K was the number of cuts that

9
had been rated as 7 plus the number of cuts that had been rated as 6, and so on until all the cuts
of each type had been cumulatively summed. The idea behind this approach is that a brain region
which responds to cuts with shorter timescales, as sensory cortices are expected to do, should
also respond to cuts where longer amounts of time pass. However, brain regions that respond
better to longer time shifts should not respond to the shorter ones. Therefore we tested K values
that include cumulatively more cuts as we moved from longer timescales to shorter ones. The
number of segments (8) the film had originally been separated into was also included as a K
value, allowing us to determine whether the segments had been successfully stitched together or
if those points of connection were showing up as additional neural state boundaries.
To test the success of the HMM we employed the principle that the neural activity of TRs
within an event should be more strongly correlated than TRs compared across event boundaries.
Each timeseries was shifted by 5 seconds and a spatial correlation was computed between each
TR and the TR of the shifted timeseries. If both TRs were within the same event, this was
classified as a “within” correlation. If the two TRs were from different events, this was
considered an “across” correlation. Using this method we calculated the average within-across
correlation.
Fitting each of the models ultimately yielded a set of within-across correlations for each
participant and each K value, which were then averaged across participants in order to identify
the K with the highest within-across difference.
Statistical thresholding: The best-fitting K value for each ROI was then tested for
significance by running a series of permutation tests. For each K value tested, an equivalent
number of K event boundaries in the data were randomly placed in the neuroimaging data for
that particular ROI. The within-across difference was then calculated for this randomly

10
constructed event structure. This process was repeated 1000 times in order to create a null
distribution of within-across differences, allowing us to compare the event structure obtained by
the HMM to others that could have appeared just by chance.

11
Chapter Four: Results
Annotations
In total, there were 865 cuts identified across the stitched together version of the 8 film
segments. Of these, 631 involved no shift in time. Amongst the cuts where a non-zero amount of
time passed between shots, the most frequent type of time shift was depicted involved
transitioning forward or backward by over one year. Figure 1 depicts a breakdown of the number
of cuts categorized in each bracket of time shift.
Fig 1. Number of cuts per time shift type, excluding cuts where no time passed between shots.
Across segments, on average each segment had 29.25 cuts with a non-zero time shift.
While the total number of non-zero cuts per segment was largely consistent across segments, a
look at the number of cuts of each time shift type within each segment revealed high amounts of
variability, with certain time shift types not present in some segments at all.
Hidden Markov Model

12
The K value that best fit both the visual and auditory cortex was 865 (average withinacross correlation visual = 0.3599530; auditory = 0.3081952), corresponding to the total number
of cuts identified in the entire film stimulus. This suggests that these lower level sensory cortices
were responsive to any and all moment-to-moment changes in the stimulus, regardless of how
much time passed between cuts. Figure 2 shows the average within-across correlation identified
by the HMM across all participants for each K-value for the visual (a) and auditory (b) cortices,
respectively.
Fig 2. Average within-across difference determined by the HMM for each K value tested for the (a) visual and (b)
auditory cortices. The green dot indicates the best fitting K value, which is the one with the highest average withinacross difference.
For the neuroimaging data obtained from ROIs located within the DMN, most of them
were best fit by a K value below the total number of cuts in the stimulus. The left angular gyrus
was best fit by a K value of 177 (average within-across correlation = 0.3237468), corresponding
to all cuts involving a time shift longer than one hour. The right angular gyrus had a larger best
fitting K value of 234 (average within-across correlation = 0.3084890), which aligned with the
total number of cuts containing a timeshift of any nonzero length. See Figure 3 for the average
within-across correlations for each K value tested by the HMM for the left (a) and right (b)
angular gyri.

13
Fig 3. Average within-across difference determined by the HMM for each K value tested for the (a) left and (b) right
angular gyrus. The green dot indicates the best fitting K value, which is the one with the highest average withinacross difference.
The mPFC had the lowest fitting value of the six ROIs that were tested, with a K of 143 (average
within-across correlation = 0.3148652). This K value equaled the number of cuts involving a
time shift longer than one day. Figure 4 shows the average within-across correlations of each K
value for the mPFC.
Fig 4. Average within-across difference determined by the HMM for each K value tested for the medial prefrontal
cortex. The green dot indicates the best fitting K value, which is the one with the highest average within-across
difference.

14
Of the ROIs tested from within the DMN, only the PMC had a K value equivalent to the
sensory cortices. The PMC was also best fit by a K value of 865 (average within-across
correlation = 0.3598219).
Fig 5. Average within-across difference determined by the HMM for each K value tested for the medial prefrontal
cortex. The green dot indicates the best fitting K value, which is the one with the highest average within-across
difference.
Permutation tests (n = 1000) were run on the best fitting K value for each ROI to test if
the within-across difference was greater than that expected by chance. All of them were found to
be statistically significant (p < 0.0001) compared to a null distribution of randomly segmented
data sets of the same K value. Figure 6 depicts the results of all the permutation tests.

15
Fig 6. Results of permutation tests run on the best fitting K value for each ROI.
Top row, from the left: mPFC, left AG, right AG.
Bottom row, from the left: PMC, auditory cortex, visual cortex.

16
Chapter Five: Discussion
The results of this study provide evidence that the DMN is less sensitive to fleeting
events than the sensory cortices. As longer narrative time shifts typically involve greater changes
in the meaning of the story, we anticipated that areas involved in narrative comprehension such
as the DMN (Jääskeläinen et al., 2021) would overlook cuts involving little to no time shift, as
compared to lower order regions of the brain. As was initially predicted, neuroimaging data from
regions in the visual and auditory cortex were found to be best fit by a number of event
boundaries equivalent to the total number of cuts in the film stimulus. This finding suggests that
these areas responded indiscriminately to moment-to-moment changes in sensory information
across all cuts, regardless of narrative time shift. These outcomes echo similar effects observed
in other work, where regions in the early auditory cortices have been shown to respond to
shifting sounds inconsiderate of temporal and narrative significance (Lerner et al., 2011).
In contrast, regions located in the DMN tended to be best fit by lower numbers of event
boundaries, aligning with expectations that the DMN would attend more strongly to cuts
involving longer time shifts. The generally lower number of discrete events identified regions of
the DMN provides preliminary support to suggest that the DMN is less sensitive to cuts lacking
meaningful changes in the narrative content and, by extension, that those meaningful changes
would align with longer time shifts. We would anticipate that cuts with longer narrative time
shifts would involve more effortful reconciliation of information than scenes where shorter time
shifts occur. Given that the DMN has been found to be involved in predictive processing (Addis,
2020) and schema representation (Gilboa & Marlatte, 2017), it follows that these moments
involving more dramatic narrative shifts would likewise evoke more activity in the DMN. These
findings would also align with prior work showing the DMN’s role in meaning-making

17
(Yeshurun, Nguyen, & Hasson, 2021) as well as temporal tracking of a scrambled narrative
(Simony et al., 2016). Taken together, these results demonstrate that the DMN might be involved
in stitching together events that are discontinuous in time to form a continuous narrative.
These results also demonstrated further evidence for a hierarchy of temporal processing
in the brain, such that the DMN appears to be processing fewer events on a longer time scale in
comparison to the lower level sensory cortices. Prior work has showcased similar effects
comparing a hierarchy of temporal processing of audio narratives in the DMN and in the early
auditory cortex (Lerner et al., 2011). The DMN has also been thought to be capable of keeping
track of episodic narrative information for up to several real-time minutes (Chen et al., 2016).

18
Chapter Six: Limitations and Future Directions
Given that this study was conducted on pre-existing data, there were several design
constraints built into this project. With only a sample size of 15 participants in the original
studyforrest audio-visual study, this necessarily limits the scope of possible analyses that can be
conducted and tempers the claims that can be elicited from our findings. Moreover, the film itself
was originally shown to the participants in 8 segments, with one long break in the middle. It’s
possible that the researchers’ enforced segmentation of what was originally a continuous film has
already had some qualitative impact on the participants’ cognitive processing of this narrative.
And while the results presented here do identify a best fitting number of event boundaries
for multiple regions of the DMN and sensory cortices, it’s still unclear whether the event
boundaries identified in the model align with the temporal location of the cuts identified by the
annotation coders. It also remains unclear whether and how the amount of time passing between
cuts impacts event processing in various regions of the brain.
Finally, due to how much data there is and how computationally demanding these
analyses were, this necessitated shrinking the size of our ROIs quite small. However, both the
size and shape (8mm spheres, 257 voxels) are ultimately quite arbitrary, and given how small
they are relative to the size of the entire region, it’s possible that these analyses would have
greater accuracy had these ROIs been placed in a different location.
Future directions
While the results presented here provide early evidence that the DMN may be involved in
tracking narrative discontinuity, follow up analyses on this data aim to further elucidate the
DMN’s particular role in temporal processing of time shifts between events. Subsequent analyses
will look at whether the location of the event boundaries identified by the HMMs align with

19
actual cuts in the film, as identified by the annotation coders. In the event that there is any
alignment between model-identified event boundaries and actual film cuts, future work will also
delve into which cuts line up for each region of interest.
Additional analyses will also explore whether there are consistent trends in the amount of
time passing between cuts that arise across brain regions. The findings obtained from the current
study showed that best fitting K values varied across regions of the DMN, suggesting that
specific brain areas within the DMN may attend differentially to longer and shorter narrative
time shifts.
Finally, it’s also possible that the direction of time shift could have some bearing on
temporal comprehension. There may, for example, be something cognitively distinct about
stitching together an event that occurs several years in the past, as opposed to one that occurs
several years in the future. In that vein, follow up analyses will explore the role of direction on
the tracking of time shifts between cuts. More broadly, subsequent work intends to understand
more about what is happening at the event boundaries themselves. Additional analyses will also
be aimed at linking the certainty of the HMM model at each boundary to the temporal properties
of each transition.

20
References
Addis, D. R. (2020). Mental time travel? A neurocognitive model of event simulation. Review of
Philosophy and Psychology, 11(2), 233-259.
Baldassano, C., Chen, J., Zadbood, A., Pillow, J. W., Hasson, U., & Norman, K. A. (2017).
Discovering event structure in continuous narrative perception and memory. Neuron,
95(3), 709-721.
Baldassano, C., Hasson, U., & Norman, K. A. (2018). Representation of real-world event
schemas during narrative perception. Journal of Neuroscience, 38(45), 9689-9699.
Chen, J., Honey, C. J., Simony, E., Arcaro, M. J., Norman, K. A., & Hasson, U. (2016).
Accessing real-life episodic information from minutes versus hours earlier modulates
hippocampal and high-order cortical dynamics. Cerebral cortex, 26(8), 3428-3441.
Field, S. (2005). Screenplay: The foundations of screenwriting. Delta.
Gilboa, A., & Marlatte, H. (2017). Neurobiology of schemas and schema-mediated memory.
Trends in cognitive sciences, 21(8), 618-631.
Hanke, M., Adelhöfer, N., Kottke, D., Iacovella, V., Sengupta, A., Kaule, F. R., ... & Stadler, J.
(2016). A studyforrest extension, simultaneous fMRI and eye gaze recordings during
prolonged natural stimulation. Scientific data, 3(1), 1-15.
Haxby, J. V., Guntupalli, J. S., Connolly, A. C., Halchenko, Y. O., Conroy, B. R., Gobbini, M. I.,
Hanke, M. & Ramadge, P. J. (2011). A Common, High-Dimensional Model of the
Representational Space in Human Ventral Temporal Cortex. Neuron, 72, 404–416.
Jääskeläinen, I. P., Sams, M., Glerean, E., & Ahveninen, J. (2021). Movies and narratives as
naturalistic stimuli in neuroimaging. NeuroImage, 224, 117445.
Lee, H., Bellana, B., & Chen, J. (2020). What can narratives tell us about the neural bases of
human memory?. Current Opinion in Behavioral Sciences, 32, 111-119.
Lerner, Y., Honey, C. J., Silbert, L. J., & Hasson, U. (2011). Topographic mapping of a
hierarchy of temporal receptive windows using a narrated story. Journal of Neuroscience,
31(8), 2906-2915.
Østby, Y., Walhovd, K. B., Tamnes, C. K., Grydeland, H., Westlye, L. T., & Fjell, A. M. (2012).
Mental time travel and default-mode network functional connectivity in the developing
brain. Proceedings of the National Academy of Sciences, 109(42), 16800-16804.
Raichle, M. E. (2015). The brain's default mode network. Annual review of neuroscience, 38,
433-447.

21
Sava-Segal, C., Richards, C., Leung, M., & Finn, E. S. (2023). Individual differences in neural
event segmentation of continuous experiences. Cerebral Cortex, bhad106.
Schacter, D. L., & Addis, D. R. (2007). The cognitive neuroscience of constructive memory:
remembering the past and imagining the future. Philosophical Transactions of the Royal
Society B: Biological Sciences, 362(1481), 773-786.
Simony, E., Honey, C. J., Chen, J., Lositsky, O., Yeshurun, Y., Wiesel, A., & Hasson, U. (2016).
Dynamic reconfiguration of the default mode network during narrative comprehension.
Nature communications, 7(1), 12141.
Speer, N. K., & Zacks, J. M. (2005). Temporal changes as event boundaries: Processing and
memory consequences of narrative time shifts. Journal of memory and language, 53(1),
125-140.
Yeshurun, Y., Nguyen, M., & Hasson, U. (2021). The default mode network: where the
idiosyncratic self meets the shared social world. Nature Reviews Neuroscience, 22(3),
181-192.
Zacks, J. M., & Tversky, B. (2001). Event structure in perception and conception. Psychological
bulletin, 127(1), 3.
Zacks, J. M., Speer, N. K., Swallow, K. M., Braver, T. S., & Reynolds, J. R. (2007). Event
perception: a mind-brain perspective. Psychological bulletin, 133(2), 273.
Zemeckis, R. (1994). Forrest Gump. Paramount Pictures.

Abstract (if available)

Abstract Films are rarely entirely linear and unbroken, often leaving viewers to fill in gaps and temporally rearrange content in order to keep track of an ongoing narrative. The Default Mode Network (DMN) has previously been implicated in mental time travel and the formation of event schemas, however it remains unclear how the DMN responds to variations in temporal distance during narrative time shifts. Drawing from Event Segmentation Theory and current models of predictive processing involved in narrative comprehension, we used Hidden Markov Models to examine brain activity during event boundaries for regions of the brain located within the DMN and lower level sensory cortical regions while watching Forrest Gump. Our results demonstrated early evidence that while areas in the early visual and early auditory cortices responded indiscriminately to all changes in information regardless of narrative significance, brain areas located in the DMN tended to respond more reliably to longer narrative time shifts. In addition to providing further evidence for a hierarchy of temporal processing in the brain, these findings demonstrate that the DMN may be involved in stitching together temporally discontinuous events in order to form a continuous narrative.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Decoding the neural basis of valence and arousal across affective states

PDF

The behavioral and neural correlates of music-evoked nostalgia across the lifespan

PDF

Spatial and temporal patterns of brain activity associated with emotions in music

PDF

Neural and behavioral correlates of fear processing in first-time fathers

PDF

Fathers’ social-cognitive network connectivity: associations with prenatal father-infant attachment and postpartum parenting behavior

PDF

Scaling control synthesis and verification in autonomy using neuro-symbolic methods

PDF

Effects of online choir and mindfulness interventions on auditory perception and well-being in older adults during the COVID-19 pandemic

PDF

Neurobiological correlates of fathers’ transition to parenthood

PDF

Responsible AI in spatio-temporal data processing

PDF

Examining the neuroanatomical correlates of reading in developmental dyslexia

PDF

Noradrenergic mechanisms of arousal-enhanced memory selectivity

PDF

Individual differences in heart rate response and expressive behavior during social emotions: effects of resting cardiac vagal tone and culture, and relation to the default mode network

PDF

Learning lists and gestural signs: dyadic brain models of non-human primates

PDF

Vascular contributions to brain aging along the Alzheimer's disease continuum

PDF

Data-driven and logic-based analysis of learning-enabled cyber-physical systems

PDF

The behavioral and neural bases of tactile object localization

PDF

A neuropsychological exploration of low-SES adolescents’ life goals and their motives

PDF

Modeling and recognition of events from temporal sensor data for energy applications

PDF

Neurological consequences of chronic anemia

PDF

Homeostatic imbalance and monetary delay discounting: effects of feeding on RT, choice, and brain response

Asset Metadata

Creator Pan, Chelsey Xin (author)

Core Title Neural processing of temporal distance in film

School College of Letters, Arts and Sciences

Degree Master of Science

Degree Program Psychology

Degree Conferral Date 2024-12

Publication Date 12/12/2024

Defense Date 12/12/2024

Publisher Los Angeles, California (original), University of Southern California (original), University of Southern California. Libraries (digital)

Tag cognitive neuroscience,computational modeling,default mode network,narrative comprehension,neuroimaging,OAI-PMH Harvest,time perception

Format theses (aat)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Kaplan, Jonas (committee chair), Habibi, Assal (committee member), Monterosso, John (committee member)

Creator Email chelspan@gmail.com,cxpan@usc.edu

Unique identifier UC11399ERKG

Identifier etd-PanChelsey-13690.pdf (filename)

Legacy Identifier etd-PanChelsey-13690

Document Type Thesis

Format theses (aat)

Rights Pan, Chelsey Xin

Internet Media Type application/pdf

Type texts

Source 20241213-usctheses-batch-1228 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu