The cognitive benefits of a common form of childhood play, fantasy play, are yet to be fully understood. During this form of play, children may engage in fantastical thinking, which can be described as “ways of reasoning about the physical world that violate known physical principles” (Woolley, 1997, p. 993). This might involve pretending to be a wizard, pretending to live in a fairy castle, or pretending to be invisible. Some children are more fantasy-oriented in their play, and others are more reality-oriented, and this seems to be a stable individual difference throughout childhood (Woolley, 1997). Furthermore, the extent to which children engage in fantasy play may be predictive of two foundational cognitive abilities—Executive Function (Pierucci, O’Brien, McInnis, Gilpin, & Barber, 2014; Thibodeau-Nielsen, Gilpin, Nancarrow, Pierucci, & Brown, 2020) and Theory of Mind (Dore & Lillard, 2015; Taylor & Carlson, 1997).

Before discussing research on fantastical thinking and cognitive development, it is important to note that the term fantasy play is often used interchangeably in the literature with pretend play or pretense. Pretend play has been described as non-literal engagement with alternate identities or events, and playful behaviours that require some type of representation or acting-as-if (Weisberg, 2015). However, fantasy or pretend play can involve activities that vary across the wide spectrum of ontological commitment, from real to non-real or fantastical. At the real end of the spectrum are activities such as pretending to be ordinary entities (e.g., a teacher) and pretending to take part in real events (e.g., a tea party). At the non-real end are activities such as pretending to be fantastical entities (e.g., a fairy) and pretending to take part in fantastical events (e.g., casting a magic spell). In the current study, we examined the potential importance of differentiating children’s fantasy play according to the extent to which it is truly fantastical (i.e., violates known physical principles) by developing a four-point fantasy orientation (FO) scale, which will be described shortly. We subsequently used this to explore potential relations between FO and creativity in childhood.

Creativity has been defined as the ability to produce ideas that are novel or original (Wallach & Kogan, 1965). Give that fantasy play also involves an element of novelty and imagination, it is not surprising that several studies have attempted to understand potential relations between the two. One extensive review of this evidence proposed that engaging in pretend activities (‘pretend’ being used here in the broad sense and not distinguishing between representational and fantastical pretense) may be a ‘crucial engine’ of child development in some domains including language and reasoning (Lillard et al., 2013). However, the review concluded that the quality of evidence concerning relations between pretense and creative thinking was ‘not convincing’ (p. 8) owing to methodological flaws, such as non-blind experimenters, and narrow ways of operationalizing creativity. However, the opposite conclusion was drawn in another review that disregarded studies with low statistical power (Silverman, 2016). Silverman concluded that it is ‘likely true’ that pretend play (again, in the broad sense of the term, not distinguishing between representational and fantastical pretense) supports the development of creativity (p. 140).

Taking Silverman’s (2016) conclusion forward, this raises the question of why fantasy and creativity might be related. Several authors have tried to answer this question. Silverman (2016) argued that processes involved in pretend play, such as identification of one object with another and projecting symbolic actions onto objects, have a lot in common with creativity processes, including generating alternatives to reality and generating new possibilities. Lillard et al. (2013) noted Dansky and Silverman’s (1973) speculation that engaging in fantasy play (used to refer broadly to representational and fantastical play) may predispose children to a playful attitude, which may be conducive to producing novel or unusual ideas. Singer and Singer (1990) claimed that children also practise divergent thinking, a form of creativity, in their pretend play by endowing objects with multiple pretend identities over time or by developing alternate story endings. These processes, including generating alternate story endings and imaginatively manipulating event scripts (Carlson & Zelazo, 2008), may serve to provide a form of separation between a child’s thoughts and their immediate behavioural environment, which may support the generation of novel ideas (Sigel, 1970, cited in Carlson et al., 2014). Similarly, Harris (2000) and Lillard (2001) suggest that engaging in pretend play occurs at a level of representation that is separate from reality, and this may support creative thinking. Finally, engaging in pretend play (both representational and fantastical) also involves counterfactual thinking (Walker & Gopnik et al., 2013), that is, generating ideas that are counter to reality, such as pretending that there is tea in an empty teapot (Harris, 2000), or that a cup contains magic potion. This type of thinking also is required to generate original, creative ideas (Magid, Sheskin, & Schulz, 2015). Taking these arguments one step further, it could be reasoned that the more fantastical the play, the stronger the relation between play and creativity, because both high fantasy and creativity involve thinking about a broad range of possibilities that are a great distance from the real world (Runco & Pina, 2013).

To assess the extent to which children’s play is fantastical, researchers have interviewed children about their play preferences, for example, their favourite book or TV show, whether they like to take on pretend identities, and whether they have an imaginary companion (Sharon & Woolley, 2004; Taylor & Carlson, 1997). Taylor and Carlson (1997) used a binary measure of FO, with responses coded as reality oriented (playing tag or preferring to play with a toy puzzle) or fantasy oriented (playing ‘house’ or preferring to play with a ninja turtle) (see also Pierucci et al., 2014 and Richert & Smith, 2011). In this scheme, any forms of pretending or play that indicated some level of representation were categorised together, that is, the level of fantasy in the fantasy-oriented category was not taken into account. In contrast, other researchers have attempted to differentiate level of fantasy to some extent, following Singer’s (1973) guidelines. For example, Sharon and Woolley (2004) differentiated fantasy-oriented responses (e.g., monster game) from imaginative or representational responses without a clear fantastical element (e.g., playing fireman) and reality-oriented responses (e.g., checkers) (see also Boerger, Tullos, & Woolley, 2009; Dore & Lillard, 2015; Woolley, Boerger, & Markman, 2004). Similarly, Thibodeau, Gilpin, Brown, and Meyer (2016) differentiated between fantastical responses (e.g., fairies or superheroes), responses involving anthropomorphised or animated entities or games (e.g., stuffed animals or toy cars), and realistic responses (e.g., checkers or bingo). Researchers in these studies have typically summed the children’s scores on these measures to provide an overall FO score.

These methods of coding and analysis are limited, however, for two reasons. First, the three-point scale does not take into account all levels of fantasy. Second, providing an overall FO score does not allow examination of whether there are any categorical differences between, for example, children who engage in imaginative but non-fantastical play (e.g., playing doctors) and children who are truly fantasy oriented (e.g., playing witches). These two limitations were addressed in the current study.

This study extended the method of coding children’s interview responses to produce a more nuanced FO score. Another level was included that comes between representational or possible fantasy (e.g., doctors) and impossible fantasy (e.g., witches): improbable fantasy (e.g., astronauts). Improbable fantasy incorporates entities and events that are theoretically possible because they do not violate any physical laws, but are practically or socially unlikely for the majority of people living in a particular context. Examples of improbable events might include owning a pet lion, meeting the Queen, or living in a tree house. Shtulman and Carey (2007) defined these events as ones that violate contingent truths, such as that lions live in the wild, or those that are statistically rare everyday occurrences, such as meeting the Queen or living in a tree house. These can be considered as upholding scientific principles, but are somewhat fantastical in their violation of what is normal in reality.

There are conceptual differences in the metaphysical possibilities that arise when engaging in representational/possible fantasy (e.g., pretending to be doctors), improbable fantasy (e.g., pretending to be astronauts), and impossible fantasy (e.g., pretending to be witches). Engaging in impossible fantasy (and to some extent, improbable fantasy) provides an infinite number of ways of achieving one’s goals because normal restrictions imposed by metaphysical scientific principles do not apply. In contrast, representational or possible acts of fantasy play tend to conform to these principles (Harris, Kavanaugh, Wellman, & Hickling, 1993). Children with a high FO, or those who engage in more improbable or impossible fantasy, thus construct a world that is more counter-to-reality than children with a low FO (Runco & Pina, 2013). This possibility is supported by results from Subbotsky, Hysted, and Jones (2010) who found that children who watched a fantastical episode (from the film series featuring a wizard named Harry Potter) scored higher on creativity tasks than those who watched a non-fantastical, but equally engaging episode from the same film.

In the current study, children’s scores were examined in three different ways for the purposes of analysis. First, a binary system was used, categorising children as producing either reality oriented or fantasy-oriented responses to determine whether there was an effect of these two categories on creativity (Taylor & Carlson, 1997). Second, the four-point FO coding system was used to explore mode of FO on creativity, to determine whether, for example, impossible fantasy was associated with more creative responses than improbable fantasy. Finally, an average FO score was used to explore linear relations with FO and creativity. Following Silverman’s (2016) conclusions, the hypothesis was that there would be significant relations between FO and creativity.

The strength of the relation between FO and creativity may, however, be dependent on other factors, including what domain of creativity is assessed in a creative thinking task, as well as individual differences, such as age or gender. Taylor et al. (2020) found evidence suggesting a link between FO and creativity for some of their creativity tasks in 8- to 12-year-olds but not for others. They found that children who invented imaginary worlds, or paracosms, provided a more creative ending to a story stem than children who had not invented paracosms. They also drew a more creative pretend person (see also Mottweiler & Taylor, 2014). However, there was no difference between the two groups of children on a divergent thinking task (generating alternate uses for an empty milk carton), a make a ‘silly’ collage task, or on a social consequences task in which children had to explain how the world would be different if people had tails. That said, there were positive correlations between scores on three creativity tasks that involved a more social element – story stem, draw a pretend person, and social consequences. Although Taylor et al. (2020) generally did not find any gender differences, girls produced more creative story endings than boys, and age was correlated with the alternate uses divergent thinking task. These findings thus lend support to the view that creativity is domain specific rather than domain general, and that FO may be differentially related to different domains of creativity (Baer, 2012; Kaufman, 2012).

However, results from another study do not strongly support these conclusions. Subbotsky et al. (2010) examined the relations between exposure to fantasy for different domains of creative performance in 4–6-year-olds. As mentioned earlier, one group of children watched a fantastical episode from the film series featuring a wizard named Harry Potter, and another group watched an equally engaging but non-fantastical clip from the same series. Children then completed a movement creativity task involving different ways to put a plastic cup into a bin, as well as a verbal task involving generating alternate uses for a cup, and a drawing task in which they had to draw a non-existent object (e.g. an animal and a house). Subbotsky et al. (2010) found that children in the fantastical film group were more creative on all of these measures compared to children in the non-fantastical film group, lending support to the view that creativity may be domain general. While they did not examine gender differences, the 6 year-olds achieved higher creativity scores than the 4 year-olds.

Although Subbotsky et al. (2010) did not directly assess individual differences in FO, and Taylor et al. (2020) found differences between FO and some creativity tasks in older children, it seems important to explore further the potential relations between FO and different domains of creativity in early childhood, and potential differences in age and gender. In the current study, we included three tests of creativity that loosely correspond to Kaufman’s (2012) three creativity domains: scholarly, performance, and artistic. Specifically, we included a verbal divergent thinking test, a physical movement task, and a draw a pretend person task. Given the theoretical arguments for a relation between FO and creativity we might expect relations between FO and all three creativity tests. However, given the evidence presented earlier that these relations might vary by type of creative output, we alternatively might expect to see domain differences. Regarding age and gender, given that some research has identified effects of these individual variables on FO and creativity (e.g., Carlson & Taylor, 2005; Subbotsky et al., 2010; Taylor et al., 2020), we explored whether these interacted with the potential relation between FO and creativity.