The Dreamcatcher: Interactive Storytelling of Dreams

EDITOR: Mike Potel, [email protected]

DEPARTMENT: APPLICATIONS

The Dreamcatcher: Interactive Storytelling

of Dreams

Edyta Paulina Bogucka , Technical University of Munich (DE), 80333 M

€

unchen, Germany

Bon Adriel Aseniero

, Autodesk Research, Toronto, ON, M5J 1W9, Canada

Luca Maria Aiello

, IT University of Copenhagen (DK), 2300 Kùbenhavn, Denmark

Daniele Quercia

, Nokia Bell Labs, Cambridge, CB3 0FA, U.K. and CUSP King’s College London, WC2R 2LS

London, U.K.

Sleep scientists have extensively validated the continuity hypothesis, according to

which our dreams reﬂect what happens during our waking life. Yet, only a few

attempts have been made to increase the general public’s awareness about the

beneﬁts of dream analysis in better understanding and improving our daily life. We

designed “The Dreamcatcher,” an interactive visual tool that explores the link

between dreams and waking life through a collection of dream reports. We

conducted a user study with 154 participants and found a 25% increase in the

number of people believing that dream analysis can improve our daily lives after

interacting with our tool. The visualization informed people about the potential of

the continuity hypothesis to a surprising extent, to the point that it increased their

concerns about sharing their own dream reports, thus opening new questions on

how to design privacy-aware tools for dream collection.

he idea that dreams might contain hidden mes-

sages that can inﬂuence our waking life has fasci-

nated humankind since the beginning of

civilization. During the second century AD, Artemidorus

Daldianus

produced a ﬁve-volume treatise entitled

“Oneirocritica” (“The Interpretation of Dreams”), in which

he associated symbolic meanings to images and situa-

tions that frequently appeared in dreams. Since then,

sleep scientists have found abundant evidence about

the connection between daily experiences and dream-

ing, which is summarized by the so-called continuity

hypothesis: most dreams are a continuation of what we

experience during the day.

This hypothesis provides a

theoretical basis for therapy, as it can be used to raise

self-awareness, identify latent emotional states, and to

help people cope with signiﬁcant life events.

Despite the continuity hypothesis being well-stud-

ied, little effort has been made to communicate its

importance to the general public. At one end of the

spectrum of dream-related research, there are proj-

ects focused on the analysis of dream reports that

rely on scales and inventories developed by psycholo-

gists.

At the other end, there are visual representa-

tions of the content of dreams that are mostly

artistic.

5,6

To ﬁll this gap between these two extremes,

we tackle the challenge of conveying different aspects

of the continuity hypothesis in a visual form that is

informative, yet appealing to a nonexpert audience.

We did so by building a visual interface for dream sto-

rytelling that uses the visual metaphor of a familiar

cultural artifact: the Dreamcatcher (publicly available

at https://social-dynamics.net/dreams/). The visualiza-

tion allows people to explore the waking-life and

dreams of seven dreamers. To achieve that, we tapped

into data from DreamBank, a public source of dream

reports, and we built upon a recently developed algo-

rithmic tool

that uses natural language processing

(NLP) to produce an automatic analysis of dreams

0272-1716 ß 2021 IEEE

Digital Object Identiﬁer 10.1109/MCG.2021.3068794

Date of current version 3 May 2021.

May/June 2021 Published by the IEEE Computer Society IEEE Computer Graphics and Applications 105

Authorized licensed use limited to: Technische Universitaet Muenchen. Downloaded on May 10,2021 at 11:08:43 UTC from IEEE Xplore. Restrictions apply.

according to the Hall-Van de Castle inventory,

a vali-

dated and widely used dream coding scale. We dem-

onstrated the effectiveness of our visualization in

conveying the implications of the continuity hypothe-

sis in daily life through a user study involving 154 Ama-

zon Mechanical Turk (AMT) crowdworkers.

METHODS

Dream Data and Coding System

Our tool visualizes a curated collection of dreams

from DreamBank, an online repository containing over

38,000 dream reports gathered from past research on

dreams.

Dreams are annotated with their dates of

recording, which span six decades (from 1960 to 2015).

Sets of dream reports are linked to free-text descrip-

tions of the dreamers, which contain information

about their gender, age (ranging from 7 to 74), profes-

sion, and personal history.

In addition to the text and metadata available in the

database, we enriched the visualization with analytics

from the Hall-Van de Castle’s dream coding system,

standard reference for the quantitative study of

dreams. The Hall-Van de Castle system consists of

numerical scales partitioned into 10 categories; the

dream analyst parses the report to count instances of

elements belonging to those categories and uses sim-

ple formulas to combine those counts to produce a

dream proﬁle. In practice, these categories are not of

equal importance in capturing the psycho-pathological

value of the dream’s content. Dream scientists deter-

mined that the three categories of characters, social

interactions,andemotions (and their subcategories)

are the most informative ones.

In this work, we

focused on these three categories.

Characters. People, animals, and imaginary ﬁgures

who appear in the dream report. We measured the

fraction of characters who are: male and those who

are female (Male% and Female%); family members of

the dreamer (Family%); animals (Animals%); either ﬁc-

tional or dead (Imaginary%).

Interactions. Interactions among characters of

two types: friendly and aggressive. We measured the

number of friendly interactions (Friends) and the num-

ber aggressive interactions (Aggression), both divided

by the total number of characters.

Emotions. Markers of positive or negative emo-

tions in the dream report. In particular, we measured

the ratio between the number of negative emotions

and the total number of emotions expressed in the

dream (Negative emotions%).

To sp ot anomalies in the content of a dream, one

needs to compare the numerical proportions

deﬁned above against the values of a “typical”

dream report. Dream researche rs previously esti-

mated such normative proportions.

Given a mea-

sured proportion p and the correspondin g

normative proportio n p

norm

, we compared them

using Cohen ’s h, a measure of distance between

two proportions:

h ¼ð2  arcsinð

ﬃﬃﬃ

ÞÞ  ð2  arcsinð

ﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃﬃ

norm

ÞÞ: (1)

Automatic Dream Coding

Traditionally, dream coding is performed manually, which

is time consuming. To quickly score large collections of

reports, in previous work, we developed a simple NLP

tool to extract the elements of the short version of the

Hall-Van de Castle scale from text.

We assessed the

RELATED WORK

ost dream-related visualizations rely on

standard charts and word clouds and are

embedded in personalized dream logging apps used

predominantly to record dreams, tag them, and share

them with others. A few original dream-related

visualizations have been proposed in the last years, one

“scrolly telling” exploration of dream-related Google

searches,

and one which attempts to map the

imaginary space of dreams with traditional cartography

tools.

The lack of a standard visualization paradigm for

dream exploration opens the possibility to create visual

forms that are less constrained by standard genres. For

example, visaphors borrow features from one domain to

highlight crucial aspects and communicate key take-

away messages in another domain.

Reshaping

established forms to a new style and for novel purposes

proved effective in evoking surprise, stimulating

curiosity, and persuading the audience.

The nature of

dream reports lends itself to this type of visualization, as

dreams anticipate retrospection, interpretation, and

remembrance.

106 IEEE Computer Graphics and Applications May/June 2021

APPLICATIONS

Authorized licensed use limited to: Technische Universitaet Muenchen. Downloaded on May 10,2021 at 11:08:43 UTC from IEEE Xplore. Restrictions apply.

accuracy of our NLP tool using a set of dream reports

from DreamBank, which have been hand-coded by dream

experts. Using a linear combination of the negative emo-

tion score and the aggressive interaction score returned

by the tool, we extracted a dream unhealthiness score,

normalized in a range between 0 (positive dream) and 1

(nightmare). We used this measure to differentiate the

visual representation of good and bad dreams.

DREAMCATCHER DESIGN

Modes of Persuasion

The design choices of the Dreamcatcher were moti-

vated by the three modes of persuasion of the Aristo-

telian rhetoric: logos (persuasion through reasoning

and facts); ethos (speaker’s credibility); and pathos

(evoking feelings and emotions).

Logos: Coding and Keywords. To convey the prac-

tical and logical implications of the continuity hypoth-

esis—that dreams reﬂect real life—we presented

dreams in relation to the analytical dimensions of the

Hall-Van de Castle scale. We showed eight dimensions

for each dream: the ratio of the number of family

members, imaginary beings, female characters, male

characters, friends, and animals among all characters,

the percentage of aggressive interactions, and the

amount of negative emotions. We used the unhealthi-

ness scores to divide dreams into four groups: night-

mares, unpleasant, neutral, and sweet dreams.

Ethos: Effective Personas. We introduced the real-

life story of the dreamers to make appear their dream

reports more trustworthy and relatable. Existing

research in dream analysis suggests that the continuity

hypothesis is mostly predominant for certain persona

archetypes: (1) special populations with negative waking

life experiences, such as mute-deaf, blind, or trauma-

tized people; and (2) individuals with clearly distinguish-

able aspects of waking life, dreaming life, and

biographical links between them.

Following these prin-

ciples, we selected seven personas, including both indi-

viduals and groups, which represent a wide spectrum of

negative, normative, and positive examples of the conti-

nuity hypothesis during different life stages (see Table 1).

Pathos: A visual metaphor. The process of build-

ing a visaphor involves transferring selected charac-

teristics from the source domain to the data

visualization domain.

We drew inspiration from the

appearance of the dreamcatcher, an artifact originat-

ing at the Native American Ojibwe tribe that is

believed to capture dreams and ﬁlter the bad ones

out.

Our Dreamcatcher is the digital interpretation

of such an artifact. Our visaphor builds on the analogy

between physical and digital: as the traditional artifact

captures dreams, our digital version automatically

captures and codes their meanings from data.

Interaction Tasks

The Dreamcatcher allows for the following four main

tasks.

T1 Select personas of interest. The user can investi-

gate the history of different personas and check

simple hypotheses on the relationship between

their waking life and their dreams. The broad

selection of personas encourages the user to

empathize with at least some of them and

TABLE 1. Personas in the Dreamcatcher.

Persona type Persona name Characteristics Aspect of the

hypothesis

#Dreams

Characteristic life Horseplayer Middle-aged married man, factory worker and

animal lover. He plays the horses and notes down

his dreams to predict the winner.

Interests and

passions

129

Izzy Teenage girl passionate about collecting her

dreams, a normative set of waking life experiences

Daily concerns

and activities.

123

Female artist An artist in late 30 s working on paintings,

photographs and ﬁlms.

Daily activities

and experiences

118

Cross-dressing

businessman

Wall Street businessman in his 50 s, married, father

with interests in cross dressing.

Self-awareness 118

Future brides College women’s dreams that involve weddings. Life events 54

Special groups Blind dreamers People affected by complete vision loss either from

birth or for over 20 years

Senses and

fantasies.

121

War veteran Vietnam war veteran who had a very intense and

traumatic experience of the conﬂict.

Psychological

issues

114

May/June 2021 IEEE Computer Graphics and Applications 107

APPLICATIONS

Authorized licensed use limited to: Technische Universitaet Muenchen. Downloaded on May 10,2021 at 11:08:43 UTC from IEEE Xplore. Restrictions apply.

become more positive about the interpretation of

their own dreams.

T2 Filter dreams by category. The user can get a

summary of the good or bad oneiric experiences

of a selected persona.

T3 Select individual dreams. The user can review the

dream’s dimensions produced by the automated

Hall-Van De Castle’s coding tool.

T4 Read dream reports and associated keywords.The

user can investigate the connection between real

life and waking life through a subset of representa-

tive dreams. The similarity between the persona’s

dreams to those that the user might have had

could contribute to trigger empathetic responses.

Interface Design

Visual Elements

The main component of the visualization mimics the

two parts of a traditional dreamcatcher artifact. These

two parts are interactive and interdependent.

The lower part contains feathers hanging on strings

(see Figure 1). Each feather represents a single dream

report and their arrangement is chronological, the left-

most feathers showing the oldest dreams. The feathers

are encoded using a colorblind-friendly diverging color

scheme with four classes. Their colors are based on the

corresponding reports ’ unhealthin ess scores and range

from red (nightmar e) to blue (sweet dream). Additionally,

we enlarged and outlined a few selected feathers to

draw user’s attention; when clicked, these special

feathers display their representative dream reports

along with a short explana tion on which aspect of the

continuity hypothesis is valid for this persona. The con-

nection between dreams and real life was emphasized

by highlighting the keywords in a report that reﬂected

the dreamer’s real-life circumstances.

The upper part of the Dreamcatcher contains a

radar chart encoding the dimensions of the Hall-van

de Castle scale of the selected dream report (see

Figure 2). Hovering on the large beads in the hoop pro-

vides the deﬁnition of each dream dimension. The

radar chart takes different forms depending on the

selected persona. For instance, the shapes associated

with future brides or a schoolgirl recall heart motifs,

while the shape associated with a war veteran has

sharp edges (see Figure 4).

FIGURE 1. Each feather is a dream report and is color-coded

depending on the dream’s unhealthiness score. If a feather’s

report is made available, then the feather is shown with a

white bold line.

FIGURE 2. Radar charts showing how a dream report scores

on the Hall-Van de Castle’s scale.

108 IEEE Computer Graphics and Applications May/June 2021

APPLICATIONS

Authorized licensed use limited to: Technische Universitaet Muenchen. Downloaded on May 10,2021 at 11:08:43 UTC from IEEE Xplore. Restrictions apply.

Visual Narrative

We aimed at capturing the user focus toward the cen-

ter of the interface by using a dark, radial gradient

color background. The introduction of the different

aspects of the continuity hypothesis unfolds frame by

frame (see Figure 3), allowing the user to incrementally

explore the dream reports.

Frame 1: The Welcome Screen. The initial screen

shows only a rhetorical question “What can we learn

from automatically interpreting thousands of

dreams?” and a start exploration button.

Frame 2: Revealing the Personas. A screen with a

sentence explaining the continuity hypothesis in plain

English, and a list of personas to choose from.

Frame 3: The Dreamcatcher. When a person is

selected, the Dreamcatcher appears at the center of

the screen. The right part of the interface provides a

short biography of that persona. Every time the user

switches to a new persona, the Dreamcatcher’s appear-

ance changes. For instance, the feathers are rotated

randomly to give a sense of movement and encourage

serendipitous exploration. By using the buttons in the

legend, the user can switch

ON and OFF respective dream

categories to ﬁlter the data and avoid visual clutter.

Frame 4: Details-on-Demand. Clicking on a single

feather displays the dream dimensions in the radar

chart. By clicking on the enlarged feather, the user is

shown the dream report with highlighted keywords on

the right panel. By comparing the colors of the catch-

ers of different personas, the user can assess the dif-

ferences in their dream patterns (see Figure 4). For

example, the war veteran had many nightmares and

his catcher’s feathers are red, while the teenage girl’s

catcher has a variety of colors, reﬂecting a good bal-

ance among all the dream categories.

EVALUATION

The goal of our user study was to test whether the

Dreamcatcher could further the awareness of the con-

tinuity hypothesis among the general public. To this

end, we measured the extent to which our study par-

ticipants changed their minds about the continuity

hypothesis after interacting with the Dreamcatcher.

Experimental Setup

We recruited 154 participants from the AMT platform

and guided them through an experiment in 6 parts.

Part 1: Prestudy questionnaire. The participants

were asked to rate three statements on a 5-point Lik-

ert scale: i) Recalling and interpreting dream patterns

improves real life.; ii) Having people share their dreams

encourages me to interpret my own.; and iii) I am will-

ing to share my dreams for scientiﬁc purposes.

FIGURE 3. Four frames and interaction tasks in our visualization. (a) Welcome screen. (b) Panel for selecting a persona (T1).

(c) Appearance of the Dreamcatcher while ﬁltering dreams by category (T2). (d) Retrieving details on a dream report (T3, T4).

User advances the visual narrative through responsive interface elements marked with the click icon. Video available: http://

social-dynamics.net/dreams/teaser.mp4.

May/June 2021 IEEE Computer Graphics and Applications 109

APPLICATIONS

Authorized licensed use limited to: Technische Universitaet Muenchen. Downloaded on May 10,2021 at 11:08:43 UTC from IEEE Xplore. Restrictions apply.

Part 2: Experimental conditions. We split the partici-

pants equally between two experimental conditions: a

treatment condition in which they were shown the

Dreamcatcher visualization, and a control condition in

which they were shown the Control Visualization: asim-

ple text-based visualization that mimics the ways dreams

databases on the Web currently present dream reports.

Part 3: Exploration. The participants were invited

to explore the interface by going through the stories

and dreams of all the different personas.

Part 4: Test question. After completing the explo-

ration, the participants were asked to answer a “test

question” that ascertains whether they paid genuine

attention to the task.

Part 5: Poststudy questionnaire. The participants

were asked again to rate the three statements of the

prestudy questionnaire (see part 1).

Part 6: Open feedback. The participant were asked

to answer shortly to a few open questions concerning

interface usability and overall experience.

To ease the interpretation of the results, for each

question the questionnaire independently, we seg-

mented our participants based on their response into

three different segments, as done in a previous work.

› Negatively Inclined (NI): Participants who rated

a statement as “not at all” or “not that much.”

› Neutral/Weakly Inclined (NWI): Participants

who rated the statement as “partly.”

› Positively Inclined (PI): Participants who rated

the statement as “somewhat” and “totally.”

To quantify opinion shifts after the use of the inter-

faces, we measured the percentage growth rate of

each segment as a consequence of receiving a given

treatment:

DNI ¼ NI

after

 NI

before

where NI

after

is the percentage of participants who

were negatively inclinded toward a statement after

experiencing the visualization, and NI

before

is the per-

centage of participants who were negatively inclinded

toward the same statement before experiencing the

visualization. In a similar way, we computed the two

remaining percentage growth rates:

DNWI ¼ NWI

after

 NWI

before

DPI ¼ PI

after

 PI

before

We also calculated the mean attitude change per

statement, which is the mean of differences between

the self-reported attitude of all users after and before

seeing the visualization.

FIGURE 4. Dreamcatchers of three personas show negative, normative, and positive examples of dreaming patterns. (a) War vet-

eran. (b) Teenage girl. (c) Future brides.

110 IEEE Computer Graphics and Applications May/June 2021

APPLICATIONS

Authorized licensed use limited to: Technische Universitaet Muenchen. Downloaded on May 10,2021 at 11:08:43 UTC from IEEE Xplore. Restrictions apply.

Results

Before the Visualization

In both conditions, before experiencing any visualiza-

tion, 51% agreed that dream analysis could improve their

real lives, 58% were willing to have their own dreams

interpreted, and 82% were willing to share their dreams

for research purposes (see Table 2).

After the Visualization

After experiencing any of the two visualizations, our par-

ticipants changed their minds about S1 (“dream analysis

improves real life”). The “positively inclined” segment

grew by 25% among those being exposed to the Dream-

catcher, and by 20% among those being exposed to the

Control Visualization (see Table 3). More interestingly,

for both S2 (“willingness to interpret”) and S3 (“willing-

ness to share”), our participants did change their mind,

but not in the way we expected: instead of being more

willing to interpret or share their own dream reports

after experiencing the visualizations, they were less

inclined to do so, and all the more so when exposed to

the Dreamcatcher. This puzzling results was later

explained with the qualitative feedback we collected.

Table 4 describes the mean attitude change by state-

ment across Dreamcatcher and Control Visualization.

The highest mean attitude change for both treatments

wasobservedinS1(“dream analysis improves real life”).

For S2 (“willingness to interpret”), the negative inclination

is higher with Dreamcatcher than for Control. The trend

is opposite for S3 (“willingness to share”).

Free-Form Feedback

After experiencing the Dreamcatcher, some participants

took a ﬁrm stand against dream interpretation. One par-

ticipant said: “Not really into dream interpretation. Some-

times things are just best kept unknown.” Two others

added: “I do question what else they take into account

to get to know me and my personality before analyzing

my dreams”;and“The people’s lifestyle being in line with

dreams is shocking.” These comments suggest that the

Dreamcatcher made visible not only the power of dream

interpretation but also its corresponding privacy con-

cerns. They also explain why there was a 10% switch

from being positively inclined toward sharing their dream

reports (S2 and S3) to be either neutrally or negatively

inclined toward it. The control group mentioned the data

privacy issues more often than the Dreamcatcher group.

Users who performed the tasks on the website argued

that the text-based interface seem to be too simple and

theywouldliketosee“more illustrative and colorful

exampl es” and “morestoriestolearnfrom.”

As for statement S1 (“dreamanalysisimprovesreal

life”), participants tended to be more favorable after

being exposed to the Dreamcatcher, and that was

reﬂected in their qualitative feedback. Indeed, to describe

the Dreamcatcher, they used words like interesting, origi-

nal, unique, beautiful, pretty, lovely, cool, enjoyable, and

engaging

. They also found the Dreamcatcher easy to use

(“The Dreamcatcher was not intuitive at ﬁrst, but eventu-

ally became straightforward enough to ﬁgure out”); found

its dream categorization useful (“The different colored

leaves is an excellent idea”); and found its use of personas

appealing (“I like the idea and would probably browse

through an entire catalog of people that choose to put

their dreams up this way,” and “ThepartthatIlikedbest

was just reading the descriptions that the subjects

offered of their dreams.”).

CONCLUSION

The Dreamcatcher offers a new way of visualizing

salient aspects associated with dream reports, fostering

a deeper understanding of the continuity hypothesis

among the general public. Our Dreamcatcher can

inform new solutions for personalized applications for

TABLE 2. Percentage of participants in the three segments

(negatively/neutrally/positively inclined toward each statement)

before being exposed to any visualization.

Statement NI NWI PI

S1 improves real life 26% 23% 51%

S2 willing to interpret 19% 23% 58%

S3 willing to share 9% 9% 82%

TABLE 3. Percentage growth rates for the three segments

(negatively/neutrally/positively inclined toward each statement)

after our participants are exposed to either the Dreamcatcher or

the control visualization.

Statement Treatment DNI DNWI DPI

S1 “improve” Dreamcatcher 15 % 10% 25%

Control 17 % 3% 20%

S2 “interpret” Dreamcatcher 13 % 3% 10%

Control 1 % 5% 6%

S3 “share” Dreamcatcher 5% 1% 6%

Control 0% 5% 5%

TABLE 4. Mean attitude change of participants by statement.

Treatment Statistics S1 S2 S3

Dreamcatcher mean change 0.494 0.260 0.156

p-value 0.000 0.073 0.149

Control mean change 0.359 0.115 0.218

p-value 0.002 0.394 0.028

May/June 2021 IEEE Computer Graphics and Applications 111

APPLICATIONS

Authorized licensed use limited to: Technische Universitaet Muenchen. Downloaded on May 10,2021 at 11:08:43 UTC from IEEE Xplore. Restrictions apply.

dream pattern analysis. The shape of the tool makes it

easy to compare the quality of dreams, yet makes it in a

privacy-aware way. At ﬁrst glance, Dreamcatcher looks

like a wallpaper or an artistic graphic. The visualization

is, therefore, incomprehensive to external viewers, yet

understandable and meaningful for the user. In a sce-

nario in which users are allowed to create and share

their own personalized Dreamcatcher with others, the

form of the tool has the potential to evoke discussions

about sleep patterns and raise awareness on their con-

nections to well-being. Indeed, after interacting with the

Dreamcatcher, 25% of our study participants changed

their minds, ﬁnding that dream analysis could indeed

improve the understanding of their waking lives. As one

expects, the ability of extracting powerful real-life

markers from dream reports inevitably results in privacy

concerns. As such, one main area of future research is

whether it is possible to analyze and visualize dreams in

a privacy-preserving way. Our results suggest that it is

possible to build technologies that bridge the current

gap between real life and dreaming, ultimately making

our “sleeping mind” quantiﬁable.

REFERENCES

1. A. Daldianus, N. Rigault, J. J. Reiske, and J. G. Reiff,

Oneirocritica, vol. 2. Warszawa, Poland: Palala Press, 2015.

2. P. McNamara, “ The Continuity Hypothesis of Dreams:

A More Balanced Account.” Sep. 2014. Accessed: Mar.

2021. [Online]. Available: https://www.psychologytoday.

com/us/blog/dream-catcher/201409/the-continuity-

hypothesis-dreams-more-balanced-account

3. M. Schredl, “Continuity in studying the continuity

hypothesis of dreaming is needed,” Int. J. Dream Res.,vol.

5, no. 1, pp. 1–8, 2012, doi: 10.11588/ijodr.2012.1.9306.

4. G. W. Domhoff, Finding Meaning in Dreams. A

Quantitative Approach. New York, NY, USA: Springer, 1996.

5. C. M. Iosifescu Enescu and L. Hurni, “Fictional

volunteered geographic information in Dream

Cartography,” Int. J. Cartography, vol. 3, no. 1, pp. 76–87,

2017, doi: 10.1080/23729333.2017.1301627.

6. M. Blagrove and J. Lockheart, “Gallery of Dreams and

Artworks.“ 2019. Accessed: Mar. 2021. [Online].

Available: https://dreamsid.com/gallery-of-dreams-

and-artworks.html

7. A. Fogli, L. M. Aiello, and D. Quercia, “Our dreams, our

selves: Automatic analysis of dream reports,” Roy. Soc.

Open Sci., vol. 7, no. 8, 2020, Art. no. 192080,

doi: 10.1098/rsos.192080.

8. F. Fragapane, “The shape of dreams. A visual

exploration of Google searches for the interpretation of

dreams,” Jan. 2020. Accessed: Mar. 2021. [Online].

Available: https://the-shape-of-dreams.com

9. D. J. Cox, “Metaphoric mappings: The art of

visualization,” in Aesthetic Computing, P. A. Fishwick, Ed.

Cambridge, MA, USA: MIT Press, 2006, ch. 6, pp. 89–114.

10. F. B. Vi



egas and M. Wattenberg, “Artistic data

visualization: Beyond visual analytics,” in Online

Communities and Social Computing, D. Schuler, Ed.

Berlin, Germany: Springer, 2007, pp. 182–191,

doi: 10.1007/978-3-540-73257-0_21.

11. A. Schneider and G. W. Domhoff, “The DreamBank,”

2020. Accessed: Mar. 2020. [Online]. Available: http://

dreambank.net/

12. C. S. Hall and R. L. Van de Castle, The Content Analysis

of Dreams. New York, NY, USA: Appleton, 1966.

13. W. Domhoff, The Scientiﬁc Study of Dreams: Neural

Networks, Cognitive Development, and Content

Analysis. Washington, DC, USA: American

Psychological Assoc., 2003.

14. K. Harris, “The legend of the dream catcher,” 2019.

Accessed: Mar. 2021. [Online]. Available: https://

historydaily.org/the-legend-of-the-dream-catcher

15. T. L. DeCicco, “What is the story telling? Examining

discovery with the storytelling method (TSM) and

testing with a control group,” Dreaming, vol. 17, no. 4,

pp. 227–238, 2007, doi: 10.1037/1053-0797.17.4.227.

16. A. V. Pandey, A. Manivannan, O. Nov, M. Satterthwaite,

and E. Bertini, “The persuasive power of data

visualization,” IEEE Trans. Vis. Comput. Graph., vol. 20,

no. 12, pp. 2211–2220, Dec. 2014, doi: 10.1109/

tvcg.2014.2346419.

EDYTA PAULINA BOGUCKA is currently a Doctoral Candi-

date with the Technical University of Munich, Munich,

Germany. She was a data visualization intern with Nokia Bell

Labs. She is the corresponding author of this article. Contact

her at [email protected].

BON ADRIEL ASENIERO is currently a Senior Research Sci-

entist with Autodesk Research, San Rafael, CA, USA. He was

a data visualization intern with Nokia Bell Labs. Contact him

at [email protected].

LUCA MARIA AIELLO is currently an Associate Professor

with the IT University of Copenhagen, Copenhagen, Denmark.

Contact him at [email protected].

DANIELE QUERCIA is currently the Department Head at

Nokia Bell Labs, Cambridge, U.K. and a Professor of urban

informatics with Kings College London, London, U.K. Contact

him at [email protected].

Contact department editor Mike Potel at potel@wildcrest.com.

112 IEEE Computer Graphics and Applications May/June 2021

APPLICATIONS

Authorized licensed use limited to: Technische Universitaet Muenchen. Downloaded on May 10,2021 at 11:08:43 UTC from IEEE Xplore. Restrictions apply.