See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/328573477

Paracosms: The Imaginary Worlds of Middle Childhood

Article in Child Development · October 2018

DOI: 10.1111/cdev.13162

CITATIONS READS

19 3,395

5 authors, including:

Marjorie Taylor Naomi R Aguiar

University of Oregon Oregon State University
60 PUBLICATIONS 4,909 CITATIONS 15 PUBLICATIONS 167 CITATIONS

SEE PROFILE SEE PROFILE

Emilee Naylor
University of Utah
3 PUBLICATIONS 32 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

The imaginary companions created by children who have lived in foster care View project

Mindfulness VS. CBT Randomized Clinical trial for opioid-treated Chronic Lower Back Pain View project

All content following this page was uploaded by Naomi R Aguiar on 01 November 2018.

The user has requested enhancement of the downloaded file.

 Child Development, xxxx 2018, Volume 00, Number 0, Pages 1–15

Paracosms: The Imaginary Worlds of Middle Childhood

Marjorie Taylor and Candice M. Mottweiler Naomi R. Aguiar
University of Oregon University of Wisconsin–Whitewater

Emilee R. Naylor Jacob G. Levernier

University of Utah University of Oregon

The invention of imaginary worlds (“paracosms”) is a creative activity of middle childhood that has previ-
ously been investigated primarily with retrospective adult reports and biographical accounts. In descriptions
collected from 8- to 12-year-old children, the prevalence was 16.9% in Study 1 (n = 77) and 17.4% in Study 2
(n = 92). Children with and without paracosms did not differ in verbal comprehension, divergent thinking
(Studies 1 and 2) or working memory (Study 2). However, children with paracosms had more difficulty with
inhibitory control (Study 2) and had higher creativity scores on a story-telling task (Studies 1 and 2). Para-
cosms provided a vehicle for stories associated with imaginary companions and/or for developing complex
narratives alone or with friends.

Story telling is a universal and early developing & Levernier, 2015). These sources indicate that
human activity that is believed to promote insight paracosms are often highly elaborated with, for
into behavior, contribute to empathy, help us find example, their own governments, geographies,
meaning in life events, and extend our experience languages, cultures, and associated artifacts. Root-
beyond personal circumstance (Bruner, 1986; Green Bernstein and Root-Bernstein (2006) have specu-
& Donahue, 2009; Nelson, 1989). There are many lated that this type of storytelling might be an early
types of childhood narrative activities, differing in sign of exceptional creativity or might set the stage
the extent to which they are social or private, based for later creative endeavors by providing “an early
on fictional or real experiences, acted out in pretend apprenticeship in absorption and persistence, dis-
play or communicated solely through language, covery, synthesis, and modeling” (p. 421). Perhaps
and focused on character or plot development the production of a paracosms develops from ear-
(Engel, 2005). lier pretend play involving imaginary companions,
In this research, we investigated an impressive, an activity that has been associated with advanced
but understudied, narrative activity associated with social understanding (Lillard & Kavanaugh, 2014;
middle childhood—the spontaneous production of Taylor & Carlson, 1997), as well as creative story
imaginary societies or worlds known as “para- telling (Mottweiler & Taylor, 2014; Trionfi & Reese,
cosms” (Silvey & MacKeith, 1988). Most of what is 2009). We explored these possibilities in two studies
known about paracosms comes from adult retro- in which the children themselves were interviewed
spective descriptions (Cohen & MacKeith, 1991; about paracosms, instead of relying on what a
Root-Bernstein & Root-Bernstein, 2006), historical sample of adults could remember about their
accounts (Root-Bernstein, 2014), and case studies childhoods.
(Root-Bernstein, 2014; Taylor, Mottweiler, Naylor,
Past Research on Paracosms
This work was supported by a grant from the Templeton
Foundation to Marjorie Taylor and a Graduate Research Fellow- The early literature on paracosms mostly
ship from the National Science Foundation to Candice Mot-
tweiler. The opinions expressed in this publication are those of includes accounts in the biographies and autobi-
the authors and do not necessarily reflect the views of the John ographies of individuals who later became famous
Templeton Foundation. We are grateful for the help of Sarah for their creative work. Root-Bernstein (2014) has
Amini, members of the UO Imagination Lab, and all the families
who participated in this study.
Correspondence concerning this article should be addressed to
Marjorie Taylor, Department of Psychology, 1227 University © 2018 Society for Research in Child Development
of Oregon, Eugene, OR, 97405. Electronic mail may be sent to All rights reserved. 0009-3920/2018/xxxx-xxxx
[email protected]. DOI: 10.1111/cdev.13162
2 Taylor, Mottweiler, Aguiar, Naylor, and Levernier

provided a comprehensive review of such historical are believed to be most common (8–12 years) were
examples, including descriptions of paracosms cre- asked about them, possibly because Cohen and
ated by Emily, Charlotte, Anne, and Branwell MacKeith (1991) believed paracosms to be rare.
Bront€e, Robert Louis Stevenson, J. R. R. Tokien, C. However, this belief was based primarily upon ret-
S. Lewis, Desmond Morris, Friedrich Nietzsche, and rospective data from adults who responded to
others. These accounts establish the substantial vari- magazine advertisements—not the most reliable
ability in the characteristics of paracosms and sug- source of information about childhood activities.
gest the possibility of a link with adult creativity. Root-Bernstein and Root-Bernstein’s (2006) esti-
Robert Silvey launched empirical research on this mated the prevalence to be 3%–12% in the general
topic by advertising widely in British newspapers population and 5%–26% in the sample of
and was able to collect descriptions of paracosms MacArthur Fellows, suggesting that although the
from 57 individuals (almost all adults; Cohen & creation of an imaginary world is not widespread,
MacKeith, 1991; Silvey & MacKeith, 1988). The it is not limited to a few rare individuals.
paracosms in this sample varied in content (e.g., the The ranges in Root-Bernstein and Root-Bern-
extent to which they were magical or naturalistic) stein’s (2006) estimates reflect their recognition of
and whether or not they were based on toy or methodological limitations in their study. Their
object props or entirely in the minds of the creators. samples were self-selected; a minority of the indi-
They appeared to serve as vehicles for story telling viduals who were sent queries about imaginary
and as a way to explore real-life interests (e.g., two worlds elected to participate (i.e., 90 of 505
boys invented “Possumbulese” as the language spo- MacArthur Fellows and 262 of about 1,000 college
ken in “Possumbol” when they started studying students). Of the individuals who completed the
languages at school) and real-life events (e.g., a gen- questionnaires, 43% of MacArthur Fellows and 40%
eral election in Britain was followed by one in of college students initially reported that they had
“Bearland”). In this self–selected sample, the cre- invented paracosms as children, but many of the
ation of paracosms peaked at about age nine and descriptions were ambiguous or included day-
was less common after age 12. dreams or play that did not involve the creation of
Research by Root-Bernstein and Root-Bernstein an imaginary world. The final estimates were based
(2006; Root-Bernstein, 2014) underscores the poten- on coding by the authors. Despite the self-selected
tial developmental significance of paracosms (also sample, coding that was not blind to group, and
referred to as “worldplay”). They reported that issues with retrospective data, Root-Bernstein and
childhood paracosms were more common for recip- Root-Bernstein’s results indicate that paracosms are
ients of MacArthur Awards than for a comparison less rare than Cohen and MacKeith (1991) believed.
group of college students, suggesting that highly To get a more accurate estimate of prevalence, we
creative people were more likely to have them. interviewed a relatively large sample of children in
Many of the participants assigned importance to the age range when paracosms are believed to peak.
having a childhood paracosm and saw it as being In addition to asking about paracosms, we included
connected to their work as adults (e.g., developing a variety of other measures as an exploratory investi-
the capacity to consider alternative possibilities). gation of the characteristics that might distinguish
Although Cohen and MacKeith (1991) speculated children who create paracosms from other children,
that paracosms might be particularly likely for indi- including assessments of creativity. The children and
viduals who later pursued artistic careers, the their parents were asked about imaginary compan-
MacArthur Fellows who reported having para- ions as well as paracosms because it seemed possible
cosms represented a wide range of disciplines, a that the creation of imaginary companions leads to
result that is consistent with Root-Bernstein and the invention of worlds for them to live in. Note,
Root-Bernstein’s (2006) interdisciplinary view of cre- however, that both Cohen and MacKeith (1991) and
ativity; “imagination and make-believe exercise Root-Bernstein (2014) are careful to distinguish
general, not specialized, skills that are relevant to between imaginary companions and paracosms,
pursuits across the arts, humanities, social sciences, describing paracosms as a later developing and less
and sciences” (p. 421). common type of imaginative activity that, unlike
play with imaginary companions, usually involves
extensive documentation (e.g., written histories,
Current Studies
maps, etc.). Perhaps the content of these imaginative
To date, there are no previous large-scale studies activities are distinct, although both might reflect an
in which children in the age range when paracosms interest in story telling.
 Paracosms 3

STUDY 1 out questionnaires in an adjacent room. The chil-

dren were interviewed about paracosms and imagi-
Our goal was to investigate the prevalence of para-
nary companions, and completed five laboratory
cosms in children aged 8–12 years, determine if the
creativity tasks and assessments of coping strategies
creation of paracosms was related to having imagi-
and verbal comprehension.
nary companions, and investigate the relation
between having a paracosm and laboratory mea-
sures of creativity. The selection of additional mea- Paracosm interview
sures was motivated in part by findings in the
One challenge for this type of study is that imag-
literature on imaginary companions suggesting that
inary worlds tend to be multifaceted and idiosyn-
children with imaginary companions endorse more
cratic, differing dramatically in their content,
positive forms of coping (Gleason & Kalpidou,
structure, and associated artifacts (Taylor et al.,
2014; Taylor, Hulette, & Dishion, 2010) and in some
2015). An interview that included questions about
studies score higher on measures of social under-
all the possible topics that are explored in para-
standing and verbal comprehension (e.g., Taylor &
cosms would be too lengthy and would include too
Carlson, 1997). We included a measure of autism
many questions that were irrelevant for a particular
spectrum disorder (ASD) because we speculated
child’s creation. Here we used an interview strategy
that children with paracosms, like children with
developed in a case study by Taylor et al. (2015)
imaginary companions, might be more socially
that was designed to allow for the efficient collec-
skilled and thus have fewer of the characteristics
tion of information about the level of detail, scope,
associated with ASD. However, some descriptions
and variation in paracosms, as reported by the chil-
of paracosms have features that resemble activities
dren themselves. The experimenter introduced the
associated with ASD (e.g., preference for solitary
topic of paracosms by saying, “Some kids your age
activity, obsessive interest in systematizing, etc.), so
tell us they have a special imaginary place that they
it was possible that children with parasosms might
think about a lot. Is that something that you like to
score higher on this measure.
do?” If the child said “no,” the experimenter asked,
“What about when you were younger, when you
were a little kid; did you ever use to think about an
Method imaginary place?” If the child said yes to either
Participants question, he or she was shown a list that included
43 topics (e.g., geography, languages, etc.) that have
Seventy-seven children 8–12 years of age been identified in past research as featured in some
(Mage = 10.18 years, SD = 1.37; 37 boys, 40 girls) and children’s paracosms. The child selected the topics
their parents participated in this study. Parents iden- that were important for his or her own paracosm,
tified the ethnicities of the children as European- and the semistructured interview focused on those
American (60), Latino/Latina (4), American Indian selections.
or Alaskan Native (2), Asian American (2), Pacific
Islander (1), and multiple ethnic background (5);
three parents did not report their children’s ethnicity. Imaginary Companion Interview
The study was conducted in a university laboratory The experimenter introduced the topic by saying,
located in a small west coast city in the United States.
Seventy-four of the 77 parents reported having An imaginary friend is someone who is make-
attended college: 33.8% had bachelor’s degrees and believe; an imaginary person or animal that you
31.2% had graduate degrees (three parents did not play with, talk to, or think about a lot (Taylor,
report their level of education). The children were Carlson, Maring, Gerow, & Charley, 2004).
recruited through local birth records and flyers Sometimes an imaginary friend is completely
posted on bulletin boards. The study was described invisible and sometimes it is a toy, like a very
as research about children’s favorite activities and special stuffed animal or doll. Do you have an
creativity. Children were paid $20 for participation. imaginary friend?

If the child said no, the experimenter asked,

Materials and Procedure
“What about when you were younger—when you
Each child was seen for an hour-long session were a little kid—did you ever had an imaginary
conducted in a university laboratory. Parents filled friend?” If child indicated that he or she currently
4 Taylor, Mottweiler, Aguiar, Naylor, and Levernier

had an imaginary companion or ever had one in about the participants. For the divergent thinking
the past, the child was asked a series of questions tasks (unusual uses and consequences) in which
about the companion (e.g., name, age, gender, children generated more than one response, it was
appearance, whether it was an invisible friend or a not possible to identify if the same or different par-
personified object, the activities that the child ticipants had generated the responses. This master
engaged in with the pretend friend, and what the list was reviewed by three individuals with some
child liked and disliked about the pretend friend). expertise in the development of imagination and cre-
ativity in childhood (e.g., undergraduate research
assistants) who then scored each response/product
Laboratory Creativity Tasks
for creativity from 1 (not creative) to 5 (highly cre-
In past research, creativity has most commonly ative). Following the recommendations of Amabile,
been assessed with divergent thinking tasks such as the judges were instructed to use their own subjec-
the unusual uses task in which participants are tive criteria for creativity and were not trained.
asked to list possible uses for an everyday object However, they were told to avoid using verbal abil-
(Wallach & Kogan, 1965). Uniqueness—the number ity or how much children said as the basis for the
of uses that are not mentioned by other participants ratings. The mean score for each response was com-
—is interpreted as an index of creativity. However, puted and the dependent measure was the child’s
uniqueness scoring has been criticized because highest score.
inappropriate or bizarre responses tend to be identi- Unusual uses task—nonsocial content. This is a
fied as unique and thus inflate creativity scores (Sil- divergent thinking task in which children generate
via et al., 2008). Silvia et al. recommend using as many uses as possible for a common object
creativity judgments instead, which also have the (Guilford, 1967). The experimenter placed a milk
advantage that they can be used with many differ- carton on the table and said,
ent types of creativity tasks. Here we used Ama-
bile’s (1982) consensual assessment technique in This is a milk carton. It’s empty; I’ve poured all
which participants’ products are presented in ran- of the milk out. You might be able to use this in
dom orders and scored for overall creativity by different ways. What I want you to do now is to
“appropriate” judges (i.e., “observers who are think of all of the different ways you could use a
familiar with the domain in which the product was milk carton. What are the different uses that you
created or the response articulated,” p. 1001). can think of for a milk carton?
Another concern about the unusual uses task is
that it focuses narrowly on how physical objects Collage task—nonsocial content. The experi-
might be used or manipulated. Mottweiler and Tay- menter placed black paper, a glue stick and 144
lor (2014) found that preschool children with imagi- paper shapes (squares, rectangles, triangles, circles,
nary companions responded more creatively than stars, and moons in eight colors) in front of the
other children for laboratory tasks that involve child and said, “I want you to make an interesting,
social content, and more generally, Mouchiroud silly design by gluing these shapes onto this piece
and Lubart (2002) have argued that creativity in the of paper (Amabile, 1982).” Children were told they
social domain is a particularly important to explore could use as many or as few shapes as they wanted
and understand. Here we included three tasks that and were given 10 min to complete their collages.
had social content: a divergent thinking task in When the collage was finished, the experimenter
which children were asked to generate possible asked the child to describe the collage and took a
consequences for a fantasy scenario (i.e., people photograph of it.
having tails), a drawing task in which children Story-stem completion task—social content. This
made up a pretend person, and a story-telling task task was based on the MacArthur Story Stem Battery
in which children completed a story begun by the in which an experimenter tells the beginning of a
experimenter. The assessment of creativity also story with the use of dolls and props and then asks
included two tasks that could be completed without children to finish the story (“Show me and tell me
reference to people: the unusual uses task and a what happens now”; Emde, Wolf, & Oppenheim,
collage task in which children were asked to make 2003; Mottweiler & Taylor, 2014). We presented chil-
a design with geometric shapes. dren with a story stem that was designed to elicit a
For each task, every response/product from range of narratives that could be assessed for creativ-
every participant was listed in a randomized order ity. The experimenter placed a felt path, a small key,
to create a master list that had no information and a small doorway on the table and walked two
 Paracosms 5

small dolls down the path. (Susan and Jane were Carlson and Moses (2001), fixed orders are stan-
used to act out the story stem for female participants; dard practice for individual differences research
George and Bob were used for male participants.) because “it is critical that the individuals be
The experimenter started the story by saying, exposed to identical stimulus contexts” (p. 1035),
“Susan/George and Jane/Bob are going for a walk including the order in which stimuli are pre-
in the woods when they see something on the sented (see Carlson & Moses for discussion of
ground. Susan/George says ‘what’s this’? Jane/Bob the interpretive problems that arise if counterbal-
says, ‘It’s a key and look! There’s a door over there’. anced orders are used in this type of research).
Susan/George says, ‘Let’s see if it fits in the door’.” The test session also included an assessment of
Then the experimenter said to the child, “Now it’s moral judgment that was part of another project
your turn—what happens next?” and will not be discussed further.
Consequences task—social content. Consequences
tasks involve divergent thinking about the conse-
Parent Measures
quences of some unusual scenario (adapted from
Christensen, Merrifield, & Guilford, 1958). For this The parents completed two questionnaires asking
study, we developed a consequences task that about their children’s creation of imaginary com-
involved thinking about people. The experimenter panions and paracosms. In addition, the parents
asked the child to list “all of the ways the world filled out the Autism Spectrum Quotient: Children’s
would be different if people had tails.” Version (AQ-child; Auyeung, Baron-Cohen, Wheel-
Draw-a-pretend-person task—social content. Our wright, & Allison, 2007). Parents indicated on a
drawing task was adapted from Karmiloff-Smith’s 4-point scale from definitely agree to definitely dis-
(1990) procedure for assessing the development of agree the extent to which 50 statements described
drawing, in which children are asked to draw real their child, with higher scores indicating that the
and pretend versions of the same object. Children child had more behaviors that are characteristic of
were provided with white paper and colorful mark- children with ASD (e.g., poorer social and commu-
ers and were asked to draw “a real person” and then nication skills).
draw “an imaginary or pretend person, someone that
you make up.” After children were finished, they
were asked to describe what they had drawn. Results and Discussion
Paracosms
Children’s Coping Strategies Checklist–Revised
Thirty-three of the 77 children (42.9%) reported
This is a 54-item interview measure in which that they did not think about an imaginary place
children were asked how often (on a 4-point scale (n = 28) or said they did but could not provide any
from never to most of the time) they used a variety of information about it (n = 5). For the 44 children who
coping strategies read to them by the experimenter said they thought about an imaginary place and pro-
(Ayers, Sandler, West, & Roosa, 1996). vided a description, the interviews were reviewed by
two researchers who also read the parent question-
naires and categorized the children as describing (a)
Peabody Picture Vocabulary Test, 4th ed
a paracosm; (b) a “preparacosm” defined as a speci-
This is a standardized measure of receptive fic place, either partly or wholly imaginary, but with
vocabulary in which the experimenter says a series little or no evidence of repeated engagement with the
of words and the child indicates the picture from a place and /or not much elaborated detail; (c) a fic-
set of four pictures that best represents each word. tional place from a book, movie or video game as
originally described by the author (i.e., the child had
not elaborated the place with personal details and
Task Order
involvement to transform it into a paracosm); (d)
The tasks were administered in the following common pretend play themes (e.g., playing with
fixed order: (a) collage, (b) consequences, (c) story- dolls); (e) a real place the child had visited or would
stem completion, (d) draw-a-pretend-person, (e) like to visit in the future (e.g., Venice), or (f) other
unusual uses, (f) Children’s Coping Strategies (e.g., a dream; see Table 1).
Checklist—Revised (CCSC–R), (g) Peabody Pic- The reliability for the coding was 82% with dis-
ture Vocabulary Test (PPVT), and (h) imaginary agreements resolved by discussion. The primary
companion/paracosm interview. According to source of information was the children’s interviews,
6 Taylor, Mottweiler, Aguiar, Naylor, and Levernier

Table 1 very well, were counted if the parent provided a

Children’s Responses When Asked If They Had an Imaginary World good description. The criteria for coding children as
(Studies 1 and 2) having a personified object included one additional
Study 1 Study 2 Total condition: To differentiate personified objects from
Coded categories n = 77 (%) n = 92 (%) N = 169 (%) comfort objects, the description had to go beyond
the physical characteristics of the object (e.g., “it’s
Paracosm 13 (16.9) 16 (17.4) 29 (17.2) soft”) to include psychological details (e.g., “she lis-
Preparacosm 7 (9.1) 10 (10.9) 17 (10.1) tens to me”). Fifty of the 77 interviews were double
Pretend play 11 (14.3) 8 (8.7) 19 (11.2) coded; agreement was 90% with disagreements
Book/movie 4 (5.2) 13 (14.1) 17 (10.1)
resolved by discussion. The primary source of infor-
Real place 5 (6.5) 3 (3.3) 8 (4.7)
mation was the child reports, with the parent
Other 4 (5.2) 1 (1.1) 5 (3.3)
Child said “no” or 33 (42.9) 41 (44.6) 74 (43.8)
reports used to help corroborate or disambiguate
provided no detail information provided by the children (e.g., to help
determine if a stuffed animal should be considered
a personified object).
Thirty-six of the 77 children (46.8%; 16 boys and
with parent questionnaires serving as an auxiliary 20 girls) reported having imaginary companions: 19
source of information. Parents often do not have invisible friends (e.g., a shy orphan with pink hair
accurate information about children’s imaginative named Zoey) and 17 personified objects (e.g., a
activities, particularly after the preschool period stuffed unicorn named Alexander). Having an
(Taylor et al., 2004). imaginary companion was not related to verbal
Although Taylor et al. (2015) speculated that ver- comprehension, gender, or age, ps > .05. but was
bal ability might be related to having a paracosm related to having a paracosm, v2(1, n = 77) = 9.01,
(based on the high Peabody scores of the children p = .003, Cramer’s V = .34. Eleven of the 13 chil-
in their case study), verbal comprehension scores dren with paracosms (84.6%) also reported having
for children with paracosms (M = 119.46, imaginary companions (compared with 25 of the 64
SD = 15.3) did not differ from the scores of children children without paracosms). For seven of these
who did not have paracosms (M = 118.36, children, the imaginary companion served as a per-
SD = 1.60). Having a paracosm was not related to sonal friend and was also included in the descrip-
gender (6 of 37 boys and 7 of 40 girls had para- tion of the paracosm (e.g., the imaginary
cosms). In addition, children categorized as having companion was an inhabitant); for the other four
paracosms did not differ in age (M = children, the imaginary companions were unrelated
117.23 months, SD = 14.61) from children who did to the paracosms. Paracosms and imaginary com-
not have paracosms (M = 123.09 months, SD = panions tended to be described as temporally
16.71), t(75) = 1.18, p = .24, Cohen’s d = .13. linked: seven children said both were current (in-
The 13 paracosms varied in content, but all cluding the four children with paracosms and imag-
included information about environments (e.g., for- inary companions that were unrelated in content),
ests, lakes, caves, etc.), inhabitants (bandits, goblins, three children said both were past, and one child
zebras, hamsters, etc.), and idiosyncratic details said the paracosm was past but the imaginary com-
(e.g., fountains that spray honey). Some children panion was current.
explicitly stated that it was hard to know where to
start with their description because so much could
Creativity
be included. For example, one child initially did not
want to circle “animals” on the list of topics “be- The creativity scores for the collage (M = 2.77,
cause those are really complicated . . . the simplest SD = 0.77; Cronbach’s a = .69), draw-a-pretend-per-
one, there are three versions of it and in the sea- son (M = 2.61, SD = 0.95; Cronbach’s a = .88), and
sons, the versions vary.” consequences tasks (M = 2.99, SD = 0.90; Cron-
bach’s a = .82) were not related to verbal compre-
hension, gender, or age, ps > .05. Unusual uses
Imaginary Companions
(M = 3.55, SD = 0.77; Cronbach’s a = .85) was cor-
Children were categorized as having an invisible related with age, (r = .25, p = .03) but not verbal
friend if they said that they had one and provided comprehension or gender, ps > .05. Story-stem com-
a good description. Past imaginary companions that pletion scores (M = 2.59, SD = 1.01; Cronbach’s
the child acknowledged, but could not remember a = .89) were higher for girls (M = 2.84, SD = 1.05)
 Paracosms 7

Table 2
Relations Between Laboratory Creativity Tasks Controlling for Verbal Comprehension, Gender and Age (Studies 1 and 2)

Draw-a-pretend person Consequences (tails) Making friends Collage Unusual uses

Story stem
Study 1 .315 .343 — .08 .06
p = .009 p = .004
Study 2 .219 .378 .207 .226 .17
p = .054 p = .001 p = .069 p = .047
Draw-a-pretend-person
Study 1 — .22 — .017 .13
p = .071
Study 2 .011 .182 .17 .18
Consequences
Study 1 — — — .11 .076
Study 2 .29 .21 .30
p = .01 p = .06 p = .007
Making friends
Study 2 — — — .18 .35
p = .001
Collage
Study 1 .16
Study 2 — — — — .19

than for boys (M = 2.30, SD = 0.91), t(71) = 2.31, collage) were not significant. See Table 3 for the
p = .02; Cohen’s d = .55, but were not related to mean creativity scores for the five laboratory tasks
verbal comprehension or age, ps > .05. as a function of having a paracosm.
Table 2 shows the correlations for the five labo-
ratory measures of creativity, controlling for verbal
Coping Strategies and Autism Quotient
comprehension (PPVT 4th ed–), gender, and age.
The dependent measures used for the divergent The CCSC–R1 has two subscales that are consid-
thinking tasks (consequences and unusual uses) in ered to involve positive strategies: Active Coping
these analyses were based on the highest ratings (e.g., “think about which things are best to do to han-
according to the judges. The creativity measures dle the problem”) and Support Seeking (e.g., “you
with social content (i.e., story-stem completion, con- talked to someone who could help you figure out
sequences, and draw-a pretend-person tasks) were what to do”). These subscales were collapsed to form
correlated with each other but not with the creativ- a Positive Strategies Aggregate (M = 2.37,
ity measures that did not have social content (i.e., SD = 0.46). The other two subscales involve less pos-
collage and unusual uses tasks). The collage and itive strategies: Distraction (e.g., “you did something
unusual uses tasks were not related to each other like video games or a hobby”) and Avoidance (e.g.,
or any of the other measures. These results support “try to put it out of your mind”). These subscales
the view of creativity as a multifaceted construct were collapsed to form a Less Positive Strategies
and the need to include measures that involve the Aggregate (M = 2.28, SD = 0.43). Neither coping
social world in addition to the more standard mea- aggregate was related to verbal comprehension, gen-
sures that typically involve the manipulation of der, or age, ps > .05. Children with paracosms scored
physical objects. higher (M = 2.80, SD = 0.63, n = 13) than other chil-
dren (M = 2.42, SD = 0.51, n = 63) for Active Coping
Strategies, t(74) = 2.35, p = .02, Cohen’s d = .66.
Relations Between Paracosms and Laboratory Tests of
Scores on the Autism Quotient (M = 57.31,
Creativity
SD = 17.06; range = 22–104) were somewhat higher
Children with paracosms scored higher on the than in Auyeung et al.’s sample of children aged 4
story stem completion task and the draw-a-pretend- to 11 years (84.4% below cutoff of 76 compared
person task than children without paracosms. The with 95%). Scores were not related to verbal com-
comparisons for the consequences task and the two prehension or age; boys scored higher (M = 61.95,
tasks with nonsocial content (i.e., unusual uses and SD = 19.17) than girls (M = 53.02, SD = 13.73),
8 Taylor, Mottweiler, Aguiar, Naylor, and Levernier

Table 3 dealing with problems and were rated by their par-

Comparisons of Children With and Without Paracosms on Laboratory ents as having fewer of the characteristics associ-
Creativity Tests (Studies 1 and 2) ated with ASD.
Children Children
with without
paracosms paracosms t-Test Cohen’s d
STUDY 2
Story stem Study 2 was designed to provide a possible repli-
Study 1 3.10 (1.34) 2.4 (0.90) t(71) = 2.06 .61 cation of the paracosm findings related to preva-
p = .04
lence, gender, and verbal comprehension, and to
Study 2a 2.88 (1.41) 2.25 (1.10) t(82) = 1.97 .50
provide more information about how paracosms
p = .053
Consequences
might be associated with creativity. In Study 1,
Study 1 3.36 (.91) 2.91 (0.89) t(73) = 1.66 .19 having a paracosm was related to performance on
p = .10 creativity measures that had social content but
Study 2 3.71 (.83) 3.41 (0.78) t(90) = 1.38 .14 not on creativity tasks that involved the manipu-
p = .17 lation of physical objects. However, two of the
Pretend person three social creativity tasks in Study 1 suggested
Study 1 3.21 (0.84) 2.49 (0.93) t(75) = 2.58 .81 fantasy responses (draw-a-pretend-person task and
p = .01 consequences), whereas neither of the nonsocial
Study 2 3.21 (1.02) 2.82 (0.98) t(90) = 1.44 .15 creativity tasks was particularly conducive to
p = .15
fantasy. Perhaps children who have paracosms
Making friends
are more comfortable with or more able to gener-
Study 2 3.33 (0.88) 2.89 (0.80) t(90) = 1.96 .52
p = .053
ate fantasy content than other children. Thus, in
Collage addition to the five creativity tasks used in Study
Study 1 2.79 (0.95) 2.76 (0.74) t(75) = 0.146 .02 1, we included a new test of divergent thinking
p = .88 that was clearly social but did not involve
Study 2 3.02 (0.94) 3.07 (0.87) t(89) = 0.19 .02 fantasy.
p = .85 In this study, we added a measure of working
Unusual uses memory and two measures of inhibitory control
Study 1 3.77 (0.80) 3.51 (0.77) t(75) = 1.12 .13 (i.e., the ability to suppress thought processes that
p = .27 interfere with goal-directed behavior). Inhibitory
Study 2 3.25 (0.73) 3.32 (0.73) t(90) = 0.18 .02
control contributes to successful performance on a
p = .85
range of cognitive tasks (Zelazo et al., 2013), but
a
The story stems from eight children could not be coded because high levels of inhibitory control are not always
of a technical problem with the audio. associated with greater creativity (White & Shah,
2006). In fact, Mednick (1962) claimed that the
ability to filter extraneous information could inter-
t(75) = 2.36, p = .02, Cohen’s d = .54. Children with fere with creativity because inhibition is associ-
paracosms scored significantly lower (M = 47.47, ated with less access to remote associations,
SD = 12.87) than children without paracosms which are considered to be the building blocks of
(M = 59.31, SD = 17.19, n = 64), t(75) = 2.35, creative breakthroughs. There is some empirical
p = .02, Cohen’s d = .78. evidence that is consistent with this view. For
example, lower behavioral and cognitive inhibition
is associated with creative personality types
Summary
(Martindale, 1999), and creative achievement has
A substantial minority of the children in this been linked to lower performance on latent inhibi-
study (16.9%) were coded as having paracosms. tion tasks (Carson, Peterson, & Higgins, 2003).
These children did not score higher on the diver- Clinical research also suggests a relation between
gent thinking or verbal comprehension tasks, but creativity and low levels of inhibition (Andreason,
they produced more creative endings to stories 1987). Given the exploratory nature of our
begun by the experimenter and showed more cre- research, we included two inhibitory control mea-
ative responses when asked to draw a pretend per- sures in order to assess the extent to which chil-
son who could not exist. They also were dren with paracosms might have particular
particularly likely to endorse positive strategies for difficulty with these types of tasks.
 Paracosms 9

Method she [same as child’s gender] could do to try to

make friends?” As with the other divergent think-
Participants
ing tasks, three judges were given a list of all the
Ninety-two children between the ages of 8 and children’s responses in a random order and asked
12 years (Mage = 120.94 months; SD = 17.72; 44 to rate each response on a scale from 1 (not creative)
boys, 48 girls) and their parents participated in this to 5 (highly creative)
study. Parents identified the ethnicities of the chil-
Story-stem completion task—social content (adapted
dren as European-American (59), Latino/Latina (5),
Emde et al., 2003).
American Indian or Alaskan Native (7), Asian (11),
Consequences task—social content. This task was
and Pacific Islander (1), and multiple ethnic back-
similar to the consequences task used in Study 1
grounds (7). Two parents did not report their chil-
except that the type of tail was more clearly speci-
dren’s ethnicity. The study was conducted in the
fied (“How would the world be different if people
same location as Study 1 with a similarly well-edu-
had long, furry tails?”) because many of the chil-
cated sample recruited in the same ways. The
dren in Study 1 had asked the experimenter for
posted flyer described the study as research about
more information (e.g., “Is it long like a monkey
children’s “favorite activities, creativity, and atten-
tail?”; adapted from Christensen et al., 1958).
tion.” Children received $20 for participation.
Draw-a-pretend-person task—social content.
Children’s Coping Strategies Checklist–Revised (Ayers
Materials and Procedure
et al., 1996)
Each child was seen for a 90-min session con- Peabody Picture Vocabulary Test, 4th ed
ducted in a university laboratory while parents Digit span (Wechsler Intelligence Scale for Children, 4th
filled out questionnaires in an adjacent room. ed.)
As in Study 1, children were interviewed about
imaginary companions and paracosms, completed Working memory was assessed with two digit
a series of creativity tasks, and were assessed for span tasks in which children were asked to repeat
coping strategies and verbal comprehension. We numbers read to them by an experimenter. For
added a measure of working memory, two inhibi- each task, the procedure was terminated when
tory control tasks, and a new divergent thinking the child missed both trials within a block. Chil-
task. dren received one point for each trial that they
remembered correctly. First children were asked
Paracosm interview (Taylor et al., 2015) to recall the numbers in the same order as they
Imaginary companion interview (Taylor et al., 2004) were read and then to recall the numbers in the
Laboratory creativity tasks reverse order.
Unusual uses task—nonsocial content. In Study
1, children were asked to generate uses for a milk Flanker Task
carton, but milk cartons are often used in school art
Inhibitory control was measured with a comput-
projects and those uses were sometimes included in
erized flanker task in which children responded to
children’s responses (Guilford, 1967). In Study 2, we
target stimuli while ignoring distracters (Zelazo
changed the target object from milk carton to brick.
et al., 2013). The child was guided through the
The experimenter put a brick on the table and asked
instructions and practice trials by the experimenter.
the child to think of all of the different ways one
The task began with a plus sign in the middle of the
could use a brick. As in Study 1, three coders were
computer screen, followed by a row of seven arrows.
given a list of all the children’s responses in a ran-
The children were instructed to put their right index
dom order and asked to rate each response on a scale
finger on the right side button on a response box
from 1 (not creative) to 5 (highly creative).
and their left index finger on the left side button.
Collage task—nonsocial content (Amabile, 1982). Their task was to press the button that matched the
Making friends task—social content. This new direction of the middle arrow. Children were given
divergent thinking task involved content that was four practice trials with feedback on their perfor-
clearly social. Children were asked to “imagine that mance. Children completed 50 test trials with a short
a kid your age just moved to a new school and break in the middle of the session. They did not
wants to make friends. What are all the things he/ receive feedback for the test trials.
10 Taylor, Mottweiler, Aguiar, Naylor, and Levernier

companions, including 30 invisible friends and 21

Trail-Making Task
personified objects. As in Study 1, children with
In this task, children first connected numbered imaginary companions (M = 121.33, SD = 14.17)
dots on a page in ascending order (i.e., drew a line did not differ from children without imaginary
from 1 to 2, 2 to 3, and so on; Spreen & Strauss, companions (M = 124.59, SD = 13.17) in verbal
1998). They were instructed to work as quickly as comprehension, p > .05; however children who
they could without lifting their pen or making mis- reported having an imaginary companion were
takes. Then children completed a version in which younger (Mage = 116.77 months; SD = 18.32) than
they switched between numbers and letters (i.e., children who reported not having an imaginary
drew a line from 1 to A, A to 2, 2 to B, and so on). companions (Mage = 126.12 months; SD = 15.66),
To assess the cost of switching from numbers to let- t(90) = 2.60, p = .01, Cohen’s d = .55. In addition, 34
ters, the time in seconds children needed to com- of the 48 girls (71%) reported having imaginary
plete the Number trials was subtracted from their companions compared with 17 of the 44 boys
time for the Number/letter trials. (38.6%), v2(1) = 9.63, p = .002, Cramer’s V = .32.
As in Study 1, the majority of the 16 children
with paracosms (13; 81.3%) also reported having
Task Order
imaginary companions; the relation between hav-
The tasks were administered in the following order: ing an imaginary companion and a paracosm was
collage, consequences, story-stem completion, unusual significant, v2(1) = 5.225, p = .022, Cramer’s
uses (brick), draw-a–pretend person, making friends, V = .24. Again the paracosms and imaginary com-
imaginary companion, and paracosm interviews, flan- panions tended to be temporally linked: seven chil-
ker task, digit span, PPVT 4th ed., Coping Skills Ques- dren said both were current, two children said
tionnaire, and trail-making. both were past, three children said the paracosm
was past but the imaginary companion was cur-
rent, and one child said the paracosm was present
Parent Measures
but the imaginary companion was past. For six
Parents completed questionnaires about their children, the imaginary companion was included
children’s paracosms and imaginary companions, in the description; for the other seven children, the
and the Autism Spectrum Quotient: Children’s Ver- imaginary companion was unrelated to the
sion (Auyeung et al., 2007). paracosm.

Creativity
Results and Discussion
Creativity scores for the unusual uses (M = 3.12,
Paracosms
SD = 0.87; Cronbach’s a = .80), collage (M = 2.92;
Sixteen of the 92 children (17.4%) were catego- SD = 0.89; Cronbach’s a = .80), draw-a-pretend-per-
rized as having paracosms, a prevalence that is simi- son (M = 2.89, SD = 0.99; Cronbach’s a = .82), and
lar to Study 1 (16.9%). See Table 1 for more consequences tasks (M = 3.46, SD = 0.79; Cron-
information about children’s responses. Reliability bach’s a = .68) were not related to verbal compre-
for the two coders was 84.8% with disagreements hension, gender, or age, ps > .05. Making friends
resolved by discussion. As in Study 1, children cate- (M = 2.98; SD = 0.83; Cronbach’s a = .82) and story
gorized as having paracosms did not differ in verbal stem completion tasks (M = 2.37, SD = 1.17; Cron-
comprehension or age from children who did not bach’s a = .95) were correlated with verbal compre-
have paracosms, ps > .05; however, in this sample, hension (r = .27, p = .01; r = .24, p = .03,
girls (12 of 48, 25%) were somewhat more likely than respectively) but not with age or gender, ps > .05.
boys (4 of 44, 9%) to describe paracosms, Fisher’s Table 2 shows the correlations for the six labora-
Exact Test (two tailed), p = .06, Cramer’s V = .21. tory measures of creativity, controlling for age, gen-
der, and verbal comprehension. In this sample,
there were more relations between the measures
Imaginary Companions
that involved divergent thinking than in Study 1.
This interview was identical to Study 1. Agree- For example, whereas the unusual uses task in
ment for the two coders was 88% with disagree- Study 1 did not correlate with any other measure,
ments resolved by discussion. Fifty-one of the 92 in Study 2, unusual uses correlated with the other
children (55.4%) reported having imaginary divergent thinking tasks (i.e., consequences and
 Paracosms 11

making friends). It is possible that the target object trials (M = 0.99; SD = 0.11), with higher scores
in Study 2 (brick) worked better for this age group indexing greater difficulty with inhibitory control.
than the one used in Study 1 (milk carton), and Flanker scores did not differ for boys and girls,
thus allowed for relations based on the similar p > .05, but were negatively correlated with verbal
methodology (i.e., divergent thinking). The new comprehension, r = .29, p = .01 and age, r = .34,
making friends task was related to tasks that also p = .002. Trail-making scores did not differ for boys
involved divergent thinking rather than to tasks and girls, and were not related to verbal compre-
that were also social in content. Overall, the distinc- hension, ps > .05, but were negatively correlated
tion between social and nonsocial content in Study with age, r = .357, p = .001.
2 was not as clear as in Study 1. The two measures of inhibitory control (i.e.,
Flanker and Trail-making) were correlated
(r = .373, p = .002), controlling for age, gender, and
Relations Between Paracosms and Laboratory Tests of
verbal comprehension. Neither measure was
Creativity
related to digit span or any of the measures of cre-
Children with paracosms scored higher than chil- ativity. However, children with paracosms scored
dren without paracosms on the story-stem comple- higher (indicative of less inhibitory control) on
tion and making friends tasks. As in Study 1, there both the flanker task (M = 0.18, SD = 0.13) and the
were no differences for the two nonsocial tasks trail-making task (M = 84.62, SD = 34.41) com-
(unusual uses and collage) and the consequences pared with children who did not have paracosms
task. In contrast to Study 1, there was no difference (flanker M = 0.11, SD = 0.11; trail-making
in performance for the draw-a-pretend-person task. M = 58.21, SD = 39.27), t(80) = 2.25, p = .027,
Table 3 shows the mean creativity scores for the six Cohen’s d = .72 and, t(80) = 2.27, p = .026, Cohen’s
tasks as a function of having a paracosm. d = .61, respectively.

Coping Strategies and Autism Quotient Summary

Scores on the AQ were somewhat higher than The prevalence of paracosms in Study 2 (17.4%)
reported by Auyeung et al. (2007; M = 52.41, was similar to Study 1 (16.9%). In this study, chil-
SD = 17.27; 93.3% below cutoff of 76). AQ was not dren with paracosms did not differ from other chil-
related to age, verbal comprehension, or gender, dren on two of the three divergent thinking tasks,
ps > .05. The Positive Strategies Aggregate the collage task, or the draw-a pretend-person task.
(M = 2.22, SD = 0.50) was related to age (r = .24, However, as in Study 1, children with paracosms
p = .03) but not to verbal comprehension or gender, produced more creative endings to stories. In addi-
ps > .05. The Less Positive Strategies Aggregate tion, they were somewhat more able to generate
(M = 2.31, SD = 0.50) was not related to age, verbal creative ways to make new friends. They did not
comprehension, or gender, ps > .05. Unlike Study 1, differ from other children in verbal comprehension
children with and without paracosms did not differ or working memory, but children with paracosms
in their coping endorsements or on the AQ. were particularly challenged by tasks in which they
had to ignore distractors or inhibit a prepotent
response.
Digit Span
The digit span composite (forward digit span
Combined Analysis: Studies 1 and 2
score plus backward digit span score) was related
to age, r = .34, p = .001, and verbal comprehension, Using the combined data for measures that were
r = .315, p = .003, but not to gender, p > .05, or any identical in Studies 1 and 2 (N = 169), a logistic
of the measures of creativity, ps > .05. The compar- regression analysis was performed to predict the
ison of children with and without paracosms was creation of paracosms using the following predic-
not significant, p > .05. tors: (a) age, (b) AQ, (c) positive coping subscale of
the CCSC–R, (d) study (Study 1 or 2), (e) imaginary
companions, (f) story-stem completion, (g) draw-a-
Flanker and Trail-Making Tasks
pretend-person, and (h) collage. A test of the full
To assess children’s ability to focus on the target model against a null model was significant,
arrow and ignore distractors, scores for the conflict v2(8) = 34.82, p < .001. Not all of the predictors sig-
trials were subtracted from scores for the no conflict nificantly aided in classification; thus, backwards,
12 Taylor, Mottweiler, Aguiar, Naylor, and Levernier

stepwise elimination was used to identify a more pretending, but there were also children whose
parsimonious model. The final model yielded three paracosms and imaginary companions were unre-
positive predictors: (a) imaginary companions, lated in content. Perhaps the relation between these
B = 1.68, Wald(1) = 9.07, p = .003, semi-partial activities is based more generally on a proclivity for
r = .22, (b) story-stem completion, B = 0.44, Wald story telling. Preschool children often use imaginary
(1) = 4.18, p = .04, semipartial r = .12, and (c) draw- companions as the vehicles for stories (Taylor,
a-pretend-person, B = 0.60, Wald(1) = 5.49, p = .02, 1999), and the construction of a narrative is one of
semi-partial r = .16. Age, AQ, positive coping, col- the central features in the development of para-
lage, and whether the child participated in Study 1 cosms (Root-Bernstein, 2014). However, we do not
or Study 2 were not significant predictors, control- want to overinterpret the relation between para-
ling for all other variables. This reduced model was cosms and imaginary companions because a limita-
a significant improvement over the null model, tion of our research is that the paracosm and
v2(4) = 30.51, p < .001, accounting for 18.5% of the imaginary companion interviews were conducted in
variance in the creation of paracosms. the same session.

GENERAL DISCUSSION Characteristics of Children With Paracosms

The production of paracosms is an impressive cre- We were particularly interested in the extent to
ative activity that has been largely overlooked as a which the creativity that one might assume is asso-
source of information about children’s developing ciated with paracosms would show up in children’s
imagination and their capacity to generate complex performance on behavioral creativity tasks in the
narratives. Our participants reported a wide range laboratory. The answer was that it depended on the
of experiences and activities when asked if they task. In both studies, when asked to make a design
had an imaginary world, but a substantial minority with geometric shapes (collage), to think of uses for
of the children (16.9% in Study 1 and 17.4% in a common object (unusual uses), or to think of how
Study 2) described an imaginary place in enough the world would be different if people had tails
detail to be coded as a paracosm. As required by (consequences), children with paracosms did not
Root-Bernstein and Root-Bernstein (2006), there was differ from other children. The null result for the
“the notion of a specific place, either partly or unusual uses task is particularly noteworthy
wholly imaginary,” and “consistent repetition over because this task is very frequently used for assess-
some period of time of specific scenario” (e.g., ing creativity.
names of people and places, evidence of systemati- However, in both Study 1 and 2, children with
zation), as well as “corroborating factors” (e.g., doc- paracosms had higher creativity scores when asked
umentation, artifacts; p. 425). to think of the ending to a story (medium effect
In some ways, the creation of imaginary worlds sizes). They also reported more creative ways to
might appear similar to pretend play involving make new friends, but this result is preliminary
imaginary companions. However, we agree with because the making friends task was only used in
Cohen and MacKeith (1991) and Root-Bernstein Study 2. In Study 1, their drawings of pretend peo-
(2014) that these two types of play are distinct. ple were rated as more creative. Although this
Imaginary companions are much more common, result was not replicated in Study 2, a combined
are associated with an earlier period of develop- analysis of the two studies identified performance
ment, and typically do not involve the same on the draw-a-pretend-person task as a predictor of
amount of documentation. In addition, imaginary having a paracosm. There were no creativity tasks
companions typically serve as friends—they pro- on which children with paracosms did not score as
vide love, support and companionship (Gleason, well as other children.
2002). In contrast, the child’s role in a paracosm In addition to the draw-a pretend-person task,
often is more like that of a creator or an observer some of the other results of Study 1 and Study 2
than a participant in a social interaction. The major- were inconsistent. For example, there was no gen-
ity of the children who created paracosms also der difference in Study 1, but in Study 2, girls were
described having had imaginary companions (24 of more likely than boys to have paracosms. Inconsis-
29; 83%). For some children, creating a paracosm tency in the findings for gender are also found for
for the imaginary companion to live in appeared to imaginary companions, both in this research (i.e.,
be the next step in the elaboration of this type of no gender difference in Study 1, but more girls than
 Paracosms 13

boys had imaginary companions in Study 2) and paracosms and that those interactions were deeply
elsewhere in the literature. In general, it is often absorbing (Taylor et al., 2015).
reported that girls are more likely than boys to There were no substantial differences between
have imaginary companions (e.g., Carlson & Taylor, children with and without paracosms in verbal com-
2005; Gleason, 2004), but in many studies, there are prehension (Studies 1 and 2) or working memory
no gender differences (e.g., Singer & Singer, 1990; (Study 2). Where we did find cognitive differences
Tahiroglu, Mannering, & Taylor, 2011). Note, how- was in Study 2 with the inhibitory control tasks. For
ever, that when differences are found, the direction both tasks, children with paracosms did not perform
of the difference is consistent. To our knowledge, as well as the other children. In other studies, inhibi-
there are no studies in which more boys than girls tory control is also sometimes negatively correlated
have imaginary companions. Our interpretation is with creative behavior (e.g., White & Shah, 2006).
that gender differences for imaginary companions, On the other hand, there is evidence of a positive
and possibly for paracosms, reflect variations in relation between the development of inhibitory con-
sample characteristics and methodologies. In the trol and imagination tasks involving pretend play in
absence of a meta-analysis to settle the matter, our preschool children (e.g., Carlson, White, & Davis-
best guess is that the gender differences for these Unger, 2014). In their literature review, Carlson and
types of imaginative play are likely to be relatively White (2013) acknowledge the complexity of this
small. relation and report the existence of a subset of
There were also inconsistencies in our results for highly imaginative children who have difficulty
the Autism Quotient and Coping Assessment. In with inhibitory control. One of the challenges for
Study 1, children with paracosms endorsed more future research will be to integrate the evidence for
positive forms of coping and were rated as having a positive relation between inhibitory control and
fewer of the characteristics associated with ASD, imagination with the results suggesting the opposite
but these results did not replicate in Study 2. Simi- pattern. Our finding of a negative correlation
lar problems with replication sometimes are found between inhibitory control and paracosms needs
in the literature on imaginary companions. For replication but could be one piece of the puzzle.
example, in some studies preschool children with
imaginary companions score higher on social
understanding tasks such as theory of mind (Lillard
Limitations of This Research
& Kavanaugh, 2014; Taylor & Carlson, 1997), but
this finding does not always replicate (Davis, An important limitation of this research is that our
Meins, & Fernyhough, 2011). participants were mostly European-American chil-
Our results do not support the claim that para- dren from a middle-class well educated community.
cosms are primarily private. Root-Bernstein and In both studies, the children’s verbal comprehen-
Root-Bernstein (2006) list the privacy of the activity sion scores were higher than populations norms
as a corroborating factor for identifying paracosms, (Study 1 M = 118.55, SD = 12.9; Study 2 M =
but 12 of the 29 children with paracosms in our 122.81, SD = 13.74). Thus the prevalence of para-
research (41.4%) reported that developing their cosms in this sample might not be representative of
paracosms was a social activity shared with family children in other communities. We hope that
members or friends. In one case, the paracosm was researchers with access to more diverse populations
widely shared among a large number of children will be encouraged by our findings to include ques-
who developed its structure and narrative both tions about paracosms in their studies.
individually and in various group settings. In fact, Our inclusion of interviews about paracosms and
Cohen and McKeith (1991)—who also defined para- imaginary companions in the same test session
cosms as private activities—“were surprised when might have inflated the extent that the two activities
they discovered that, often, these were not solitary were found to be related. It is also possible that the
games. Children shared their worlds either with creativity scores in our studies might have been
brothers or sisters, or, sometimes with friends” (p. enhanced by recruitment procedures. We avoided
102). Indeed, for some children, paracosms are cen- explicit mention of “imaginary worlds” during
tral to social interactions. For example, in a qualita- recruitment, but parents’ knowledge that the
tive study of four children with paracosms, all four research concerned creativity might have resulted in
parents indicated that their children frequently a sample of children whose parents believed them
worked together on the narratives for their to be particularly creative.
14 Taylor, Mottweiler, Aguiar, Naylor, and Levernier

Conclusion Carlson, S. M., & Moses, J. L. (2001). Individual differ-

ences in inhibitory control and children’s theory of
One of the striking findings about these studies is mind. Child Development, 72, 1032–1053. https://doi.
the surprising prevalence of paracosms. Paracosms org/10.1111/1467-8624.00333
are so elaborate and inventive, it is easy to assume Carlson, S. M., & Taylor, M. (2005). Imaginary compan-
they must be rare. But in both studies about 17% of ions and impersonated characters: Sex differences in
the children described imaginary worlds—places children’s fantasy play. Merrill Palmer Quarterly, 51, 93–
where, for example, people only read the books 118. https://doi.org/10.1353/mpq.2005.0003
they have written themselves, cats fly, and inhabi- Carlson, S. M., & White, R. E. (2013). Executive function,
tants speak a language called “Plifcy.” It is also sur- pretend play, and imagination. In M. Taylor (Ed.), The
Oxford handbook of the development of imagination (pp.
prising that the children who create paracosms do
161–174). New York, NY: Oxford University Press.
not stand out more than they do. They are similar
Carlson, S. M., White, R. E., & Davis-Unger, A. C. (2014).
to their peers in verbal comprehension, working Evidence for a relation between executive function and
memory, and the most commonly used creativity pretense representation in preschool children. Cognitive
task in which children are asked to generate uses Development, 29, 1–16. https://doi.org/10.1016/j.cogde
for a common object. Where they do stand out is in v.2013.09.001
story telling. Not only were the narratives of their Carson, S. H., Peterson, J. B., & Higgins, D. M. (2003).
paracosms impressive, they invented more creative Decreased latent inhibition is associated with increased
endings to a story than the endings proposed by creative achievement in high-functioning individuals.
other children, a finding that was consistent across Journal of Personality and Social Psychology, 85, 499–506.
the two studies and in the combined analysis. https://doi.org/10.1037/0022-3514.85.3.499
Christensen, P. R., Merrifield, P. R., & Guilford, J. P.
As they mature, perhaps narrative and story-tell-
(1958). Consequences: Manual for administration, scoring,
ing will continue to be a favorite fantasy pastime
and interpretation. Beverly Hills, CA: Sheridan Supply.
for children who have paracosms, or perhaps their Cohen, D., & MacKeith, S. A. (1991). The development of
narrative skills will be reflected in the way that imagination: The private worlds of childhood. London, UK:
they think about their life goals and consider alter- Taylor & Frances/Routledge.
natives to their personal circumstances. In adults, Davis, P. E., Meins, E., & Fernyhough, C. (2011).
the extension of experience that is possible through Self-knowledge in childhood: Relations with chil-
narrative is believed to promote insight into human dren’s imaginary companions and understanding of
behavior (Bruner, 1986). Research on paracosms mind. British Journal of Developmental Psychology, 29,
could support this view, underscoring the value of 680–686. https://doi.org/10.1111/j.2044-835X.2011.
narrative activities and showing how children draw 02038.x
Emde, R. N., Wolf, D., & Oppenheim, D. (2003). Revealing
upon and extend their actual experience in the con-
the inner worlds of young children: The MacArthur Story
text of their imaginary worlds.
Stem Battery and parent–child narratives. New York, NY:
Oxford University Press.
Engel, S. (2005). The narrative worlds of what is and
References
what if. Cogntive Development, 20, 514–525. https://doi.
Amabile, T. M. (1982). Social psychology of creativity: a org/10.1016/j.cogdev.2005.08.005
consensual assessment technique. Journal of Personality Gleason, T. R. (2002). Social provisions of real and imagi-
and Social Psychology, 43, 997–1013. https://doi.org/10. nary relationships in early childhood. Developmental
1037/0022-3514.43.5.997 Psychology, 38, 979–992. https://doi.org/10.1037/0012-
Andreason, N. C. (1987). Creativity and mental illness: 1649.38.6.979
Prevalence rates in writers and their first-degree rela- Gleason, T. R. (2004). Imaginary companions and peer
tives. American Journal of Psychiatry, 144, 1288–1292. acceptance. International Journal of Behavioral Development,
Auyeung, B., Baron-Cohen, S., Wheelwright, S., & Allison, 28, 204–209. https://doi.org/10.1080/01650250344000415
C. (2007). The autism spectrum quotient: Children’s ver- Gleason, T. R., & Kalpidou, M. (2014). Imaginary com-
sion (AQ-child). Journal of Autism and Developmental panions and young children’s coping and competence.
Disorders. https://doi.org/10.1007/s10803-007-0504-z Social Development, 23, 820–883.
Ayers, T. S., Sandler, I. N., West, S. G., & Roosa, M. W. Green, M. C., & Donahue, J. K. (2009). Simulated worlds:
(1996). A dispositional and situational assessment of Transportation into narratives. In K. Markman, W.
children’s coping: Testing alternative models of coping. Klein, & J. Suhr (Eds.), The handbook of imagination and
Journal of Personality, 64, 923–958. https://doi.org/10. mental simulation (pp. 241–254). New York, NY: Psy-
1111/j.1467-6494.1996.tb00949.x chology Press.
Bruner, J. (1986). Actual minds possible worlds. Cambridge, Guilford, J. P. (1967). The nature of human intelligence. New
MA: Harvard University Press. York, NY: McGraw-Hill.
 Paracosms 15

Karmiloff-Smith, A. (1990). Constraints on representa- Spreen, O., & Strauss, E. (1998). A compendium of neu-
tional change: Evidence from children’s drawing. Cogni- ropsychological tests: Administration, norms and commen-
tion, 34, 57–83. https://doi.org/10.1016/0010-0277(90) tary (2nd ed.). New York, NY: Oxford University Press.
90031-E Tahiroglu, D., Mannering, A. M., & Taylor, M. (2011).
Lillard, A. S., & Kavanaugh, R. D. (2014). The contribu- Visual and auditory imagery associated with children’s
tion of symbolic skills to the development of an explicit imaginary companions. Imagination, Cognition and Per-
theory of mind. Child Development, 85, 1535–1551. sonality, 31, 99–112. https://doi.org/10.2190/IC.31.1-2.i
https://doi.org/10.1111/cdev.12227 Taylor, M. (1999). Imaginary companions and the children
Martindale, C. (1999). Biological bases of creativity. In R. who create them. New York, NY: Oxford University
J. Sternberg (Ed.), Handbook of creativity (pp. 137–152). Press.
New York, NY: Cambridge University Press. Taylor, M., & Carlson, S. M. (1997). The relation between
Mednick, S. A. (1962). The associative basis of the creative individual differences in fantasy and theory of mind.
process. Psychological Review, 69, 220–232. https://doi. Child Development, 68, 436–455. https://doi.org/10.
org/10.1037/h0048850 2307/1131670
Mottweiler, C. M., & Taylor, M. (2014). Elaborated role Taylor, M., Carlson, S. M., Maring, B. L., Gerow, L., &
play and creativity in preschool age children. Journal of Charley, C. M. (2004). The characteristics and correlates
Aesthetics, Creativity and the Arts, 8, 277–286. of fantasy in school-age children: Imaginary compan-
Mouchiroud, C., & Lubart, T. (2002). Social creativity: A ions, impersonation, and social understanding. Develop-
cross-sectional study of 6- to 11-year-old children. Inter- mental Psychology, 40, 1173–1187. https://doi.org/10.
national Journal of Behavioral Development, 26, 60–69. 1037/0012-1649.40.6.1173
https://doi.org/10.1080/01650250042000591 Taylor, M., Hulette, A. C., & Dishion, T. J. (2010). Longi-
Nelson, K. (Ed.). (1989). Narratives from the crib. Cam- tudinal outcomes of young high-risk adolescents with
bridge, MA: Harvard University Press. imaginary companions. Developmental Psychology, 46,
Root-Bernstein, M. (2014). Inventing imaginary worlds: 1632–1636. https://doi.org/10.1037/a0019815
From childhood play to adult creativity across the arts and Taylor, M., Mottweiler, C. M., Naylor, E., & Levernier, J.
sciences. New York, NY: Rowman & Littlefield (2015). Imaginary worlds in middle childhood: A quali-
Publishers. tative study of two pairs of coordinated paracosms.
Root-Bernstein, M., & Root-Bernstein, R. (2006). Creativity Research Journal, 27, 167–174. https://doi.org/
Imaginary worldplay in childhood and maturity and 10.1080/10400419.2015.1030318
its impact on adult creativity. Creativity Research Journal, Trionfi, G., & Reese, E. (2009). A good story: Children
18, 405–425. https://doi.org/10.1207/s15326934crj1804 with imaginary companions create richer narratives.
Silvey, R., & MacKeith, S. (1988). The paracosm: A special Child Development, 80, 1301–1313. https://doi.org/10.
form of fantasy. In D. C. Morrison (Ed.), Organizing 1111/j.1467-8624.2009.01333.x
early experience: Imagination and cognition in childhood Wallach, M. A., & Kogan, N. (1965). Modes of thinking in
(pp. 173–197). Amityville, NY: Baywood. young children. New York, NY: Holt, Rinehart, & Winston.
Silvia, P. J., Winterstein, B. P., Willse, J. T., Barona, C. M., White, H. A., & Shah, P. (2006). Uninhibited imaginations:
Cram, J. T., Hess, K. I., Martinez, J. L., & Richard, C. A. Creativity in adults with attention-deficit/hyperactivity
(2008). Assessing creativity with divergent thinking disorder. Personality and Individual Differences, 40, 1121–
tasks: Exploring the reliability and validity of new sub- 1131. https://doi.org/10.1016/j.paid.2005.11.007
jective scoring methods. Psychology of Aesthetics, Creativ- Zelazo, P. D., Anderson, J. E., Richler, J., Wallner-Allen,
ity, and the Arts, 2, 68–85. https://doi.org/10.1037/ K., Beaumont, J. L., & Weintraub, S. (2013). NIH Tool-
1931-3896.2.2.68 box Cognition Battery (CB): Measuring executive func-
Singer, D. G., & Singer, J. L. (1990). The house of make- tion and attention. Monographs of the Society for Research
believe: Children’s play and developing imagination. in Child Development, 78(4), 16–33. https://doi.org/10.
Cambridge, MA: Harvard University Press. 1111/mono.12032

View publication stats