1. Introduction
The world around fictions with imaginary worlds draw acclaim from the public, the critics, and the industry, making them both best-selling and most-appreciated fictions (e.g., top-ranked in online ranking websites). For instance, The Lord of the Rings novels are among the best-selling novels ever written, with more than 150 million copies sold in 38 different languages. The screen adaptations by Peter Jackson grossed each around 1 billion dollars in box-office worldwide, making them among the highest-grossing films ever produced. They received universal critical acclaim and won 17 Academy Awards. The Return of the King alone won 11 of them, setting the current record of the most Oscars won by a single movie. This sequel is the fifth film in the “all-time top-rated movies” list from IMDb, the biggest user-generated ranking dataset on films. Producers of fictions know just how lucrative this kind of fiction can be: After a competitive bid against HBO and Netflix, Amazon bought the rights to produce TV programs based on Tolkien's imaginary world for 250 million dollars, which is by far the most expensive script idea ever sold. Tolkien's world has also been adapted into theatre, radio, board games, video games, and role-playing games (e.g., Dungeon and Dragons and Middle-Earth Role Playing).
Several other fictions can be mentioned: Star Wars, the most successful fiction merchandising franchise of all time (Block & Wilson, Reference Block and Wilson2010) and arguably the most influential movie in the history of films (Canet, Reference Canet2016); Harry Potter, the best-selling book series in history, translated into 80 languages; Game of Thrones, whose final episode set the all-time audience record for a TV series with 16.4 million people watching it live and 15 million people streaming it later; The Legend of Zelda, one of the best-selling video game series worldwide, with over 100 million video games sold since the first one; and the Marvel Cinematic Universe, a shared imaginary world including a series of super hero films with common settings and characters, which cumulates several records in the history of cinema, including the highest opening week gross (Avengers: Endgame), the biggest opening week-end (Avengers: Infinity War) and, more significatively, the highest-grossing movie of all time (Avengers: Endgame). A more quantitative approach confirms that fictions with imaginary worlds are highly successful in recent times: the numbers of novels with imaginary worlds and the number of films with imaginary worlds have considerably increased in the last 100 years, both in absolute terms and relatively to the global levels of production of novels and films (Dubourg, Thouzeau, de Dampierre, & Baumard, Reference Dubourg, Thouzeau, de Dampierre and Baumard2021).
The examples of The Lord of the Rings, Star Wars, and Harry Potter are startling, because they are without any doubt the most popular fictions worldwide and, at the same time, the fictions which may have pushed the building of imaginary worlds the furthest, notably if we consider the amount of background information generated by their creators. Tellingly, in cultural studies, it has been argued that “more and more, storytelling has become the art of worldbuilding, as artists create compelling environments that cannot be fully explored or exhausted within a single work or even a single medium” (Jenkins, Reference Jenkins2006, p. 117; see also Besson, Reference Besson2015; Saler, Reference Saler2012; Wolf, Reference Wolf2012, Reference Wolf2017a, Reference Wolf2021). This goes to the point that imaginary worlds without any narrative have become to appear. One compelling example is the Codex Seraphinianus, an encyclopedia of an imaginary world with no story or protagonist (Serafini, Reference Serafini1981). Imaginary worlds are also invading mainstream literature with novels such as The Possibility of an Island by Michel Houellebecq which blurs the lines among science-fiction, alternative history, and highbrow general fiction.
This cultural phenomenon is not at all limited to the United States and the English-speaking Western countries. Harry Potter, Star Wars, and The Lord of the Rings are highly popular all around the world, and in fact, these franchises make most of their revenues outside North America (Kuipers & de Kloet, Reference Kuipers and de Kloet2009). Also, several imaginary worlds consumed all around the world are produced outside the English-speaking world, from the Japanese manga One Piece, which has become the best-selling manga series in history with its 470 million copies sold in 43 different countries, to Liu Cixin's best-seller The Three-Body Problem, the first Asian novel to win the American Hugo Award (Chau, Reference Chau2018). In this sense, a psychological and evolutionary understanding of the appeal for imaginary worlds is long overdue.
2. Imaginary worlds and world-dominant fictions
Fictions differ in the degree to which they distinguish themselves from the real world. For instance, in Balzac's novels, the fictional environment depicted is identical to France from the author's period and, while Balzac added approximately 3,000 fictional individuals, we intuitively picture the protagonists within the real world (Pavel, Reference Pavel2017). We infer much information derived from our folk knowledge of the real world, such as the country in which the fiction takes place, its geography, its political institutions, and the technology available. In fiction study, this idea is captured by the notion of “principle of minimal departure” (Ryan, Reference Ryan1991; Searle, Reference Searle1975) or “reality principle” (Pavel, Reference Pavel2017; Walton, Reference Walton1993). Conversely, Tolkien's “subcreated” world, Arda, is ostentatiously different from any environment in the real world (Tolkien & Tolkien, Reference Tolkien and Tolkien2006), and so is One Piece world, with its imaginary planet and its only continent called the Red Line. Both imaginary worlds are not only populated by humans, but also by several other imaginary races, and both include several elements that do not exist in the real world.
Following these examples, the key determinant of our definition of an imaginary world is the background knowledge required to understand the fiction, because it differs from the knowledge of the real world. Imaginary worlds are fictional environments that the recipients of the fiction could not have possibly explored in real life, be it far removed islands, locations in the future or the distant past, other planets, or environments in alternative history. In fact, in the example of Tolkien, the background information is commonly considered by literary critics as the most important feature (Jourde, Reference Jourde1991). In 1956, Tolkien wrote a letter which displays the crucial interest of his readers for background information:
Most people want more (and better) maps; some wish more for geological indications than place-names; many want more specimens of Elvish, with structural and grammatical sketches; others ask for metrics and prosodies, not only of the Elvish, but of the “translations” that are in unfamiliar modes – such as those composed in the strictest forms of Anglo-Saxon verse (e.g., the fragment on the Battle of Pelennor, Book Five, vi, 124). Musicians want tunes and musical notations. Archaeologists enquire about ceramics, metallurgy, tools and architecture. Botanists desire more accurate descriptions of the mallorn, of elanor, niphredil, alfirin and mallos, and of symbelmynë. (…) Historians require more details about the social and political structure of Gondor, and the contemporary monetary system (Letter to H. Cotton Minchin, April 16, 1956).
To sum up, by definition, a fiction based on an imaginary world is a fiction in which the consumer will learn a lot of novel information about the fictional environment (also called the “chronotope” in literary theory and philosophy of language; Bakhtin and Emerson, Reference Bakhtin and Emerson1984). For example, Harry Potter has Hogwarts and several other magical locations, Jules Verne's From the Earth to the Moon has the Moon (which had obviously not yet been explored at the time of the author), the Odyssey has the Cyclops Islands, Aiaia, the Fortunate Islands, the Siren Island, and the Lotophages, to name only a few invented islands, and The Lord of the Rings is set in a complex alternate world with hundreds of invented locations. It is worth noting that religious narratives also involve unknown worlds that are, in a way, imaginary, with different physical laws and spatial structures, for instance. However, religious narratives cannot be considered as fictions. In this paper, we will limit our study to fictional worlds, mostly because non-fictional worlds such as religious worlds are likely to be culturally stabilized for reasons (e.g., authority) that differ from pure entertainment (Boyer, Reference Boyer2001).
Note that fictions with imaginary worlds are not the only fictions in which the background information is central. One could also mention historical novels such as Umberto Eco's The Name of the Rose and social realistic novels such as Zola's Germinal, where the environments, be it a medieval Benedictine monastery or as industrial mine complex, play a central role in the fiction, although it resembles the real world. Thus, fictions with imaginary worlds belong to the broader category of “world-dominant fictions,” as opposed to “story-dominant fictions” (Ryan, Reference Ryan2014). It is also important to note that our understanding of fictions with imaginary worlds is very close to the category of “speculative fictions,” which encompasses any fictional genre typically containing some background elements that do not exist in the real world (Table 1). Yet it does not totally overlap with it. Indeed, not all speculative fictions require extensive background information to be appreciated. For instance, Edgar Poe's and Franz Kafka's fantastic tales in which there is only one element of supernatural do not offer much to explore. The appeal of these speculative fictions, in which the world is either relatively unimportant or very similar to the real world, would rely on the blurring of the boundaries between what is real and what is unreal (Todorov, Reference Todorov2015). Nonetheless, speculative fictions, being recorded and tagged online, can be used as a proxy for fictions with imaginary worlds (Dubourg et al., Reference Dubourg, Thouzeau, de Dampierre and Baumard2021).
Table 1. Fictional genres of the broad category of speculative fictions, conducive to the building of imaginary worlds, with the definitions from Wikipedia and some modern examples
Fantasy and science fiction are recent fictional genres, but imaginary worlds are much more ancient (Scholes & Rabkin, Reference Scholes and Rabkin1977). A huge number of ancient fictions set an imaginary world, in ancient epic poems about heroes' journeys, travelers' tales from the exploration age, adventure fictions, utopias, and dystopias (Wolf, Reference Wolf2012). For instance, the ancient Mesopotamian The Epic of Gilgamesh (1800 BCE) and The Odyssey (850 BCE) are often mentioned as precursors of fantasy fiction. Some scholars have tried to map the faraway lands and islands visited by Odysseus (Clay, Reference Clay2007). Other imaginary locations from this period were directly described without any narration (e.g., Arimaspi, the imaginary world from Herodotus' Histories). Lucian of Samosata's True History is evidence that such travelers' tales from this ancient time were considered as imaginary by their audience, as it is clearly stated in the Introduction: “I see no reason for resigning my right to that inventive freedom others enjoy […]. My subject is, then, what I have neither seen, experienced, nor been told, what neither exists nor could conceivably do so. I humbly solicit my readers’ incredulity” (Lucian of Samosata, 150 C.E.). Other ancient imaginary worlds are mentioned to strengthen this argument (Table 2).
Table 2. Examples of imaginary worlds in non-contemporary fictions, with their invented toponyms, extracted from the broadly inclusive list of imaginary worlds put forward by Wolf (Reference Wolf2012)
a Dates are approximative.
All these suggest that, in fictions, imaginary worlds are highly appealing. It raises several questions. Why this urge to create new fictional locations from scratch? The same stories could take place in faithful representations of the real world and it would considerably reduce the costs of fiction making (e.g., economic costs related to special effects in films with imaginary worlds). Why are we captivated by fictions with imaginary worlds, and seemingly more and more so? The timing of their success suggests that we are more predisposed to appreciate such fictions in modern societies, or we would have invented more imaginary worlds much earlier. Why are best-rated video games those with large open worlds (e.g., Zelda, Assassin's Creed, and No Man's Sky)? Why Baum (The Wizard of Oz), Tolkien (The Lord of the Rings), Lucas (Star Wars), Cameron (Avatar), Rowling (Harry Potter), and developers from Hello Games Studio (No Man's Sky), to name only a few, were willing to devote multiple years of their lives building extensive imaginary worlds? In a nutshell: Why imaginary worlds?
3. The psychological foundations and cultural evolution of fictions
To understand the human's interest in imaginary worlds, we first need to clarify why humans produce and consume fictions. At the proximate level, evolutionary and cognitive approaches to fictions have demonstrated that fictions tend to recycle and exaggerate the most attention-grabbing and fitness-relevant stimuli in real life (Boyd, Reference Boyd2018; Carroll, Reference Carroll2012; Gottschall, Reference Gottschall2012; Gottschall & Wilson, Reference Gottschall and Wilson2005; Mar & Oatley, Reference Mar and Oatley2008; Saad, Reference Saad2012; Schaeffer, Reference Schaeffer1999), such as romantic relationship (Alberti, Reference Alberti2013; Baumard, Huillery, & Zabro, Reference Baumard, Huillery and Zabroin press; Cox & Fisher, Reference Cox and Fisher2009; Martins & Baumard, Reference Martins and Baumard2021; Salmon & Symons, Reference Salmon and Symons2004; Vanderbeke, Reference Vanderbeke, Vanderbeke and Cooke2019), cooperation and cheating (Singh, Reference Singh2021), social status (Nettle, Reference Nettle2005a, Reference Nettle, Gottschall and Wilson2005b), and political rivalries (Jobling, Reference Jobling2001). In other words, fictions constitute “intensified stimuli” or “superstimuli” in the sense that they are crafted to artificially grab the consumers' attention, just like masks artificially trigger the human face detection capacity, and cuisine artificially triggers the nutrients detector systems (Boyer, Reference Boyer2018; Buss, Reference Buss2015; Nettle, Reference Nettle2005a; Sperber, Reference Sperber1996; Sperber & Hirschfeld, Reference Sperber and Hirschfeld2004; Verpooten & Nelissen, Reference Verpooten and Nelissen2010). Triggering people's attention is indeed the most important element of success of fictions. Such an evolutionary and cognitive approach of fiction, therefore, predicts that to answer the question “Why do people enjoy fictions?” is very close to answering the question “Why do people enjoy life?” (Bloom, Reference Bloom2010; Pinker, Reference Pinker1997).
A common view in behavioral sciences is that the capacity to tell stories is adaptive. This capacity would have evolved either to convey and teach new information (Sugiyama, Reference Sugiyama1996, Reference Sugiyama2001) or to simulate the real world (Mar & Oatley, Reference Mar and Oatley2008; Morin, Acerbi, & Sobchuk, Reference Morin, Acerbi and Sobchuk2019; Zunshine, Reference Zunshine2006). Here, we rather assume that humans did not specifically evolve the capacity to tell stories, but they rather create fictions thanks to a range of other adaptations (e.g., language, capacity to simulate, and theory of mind; Mellmann, Reference Mellmann, Gansel and Vanderbeke2012). Yet, because they are highly attractive, fictions can be used to fulfill any evolutionary relevant purpose that needs others' attention to be caught, be it signaling one's values to potential mates (Miller, Reference Miller2001) or cooperative partners (André & Baumard, Reference André and Baumard2020; André, Baumard, & Boyer, Reference André, Baumard and Boyer2020; Bourdieu, Reference Bourdieu2010; Nettle, Reference Nettle, Gottschall and Wilson2005b; Singh, Reference Singh2020; Reference Singh2021; Veblen, Reference Veblen1899), transmitting knowledge (Nakawake & Sato, Reference Nakawake and Sato2019; Schniter, Wilcox, Beheim, Kaplan, & Gurven, Reference Schniter, Wilcox, Beheim, Kaplan and Gurven2018; Sugiyama, Reference Sugiyama2021), communicating social norms (Ferrara, Banerjee, & Orozco, Reference Ferrara, Banerjee and Orozco2019; Mar & Oatley, Reference Mar and Oatley2008), or selling products (Saad, Reference Saad2012; Saad & Gill, Reference Saad and Gill2000). Thus, to use a standard term, we do not consider fictions as “by-products,” because they clearly confer benefits to both the producers and the consumers (André et al., Reference André, Baumard and Boyer2020). At the same time, fictions are definitely artificial. Hence, it is probably more appropriate to say that fictions (e.g., novels, films, and video games) are a kind of cultural technology primarily designed for entertainment (Dubourg & Baumard, Reference Dubourg and Baumardin press; Singh, Reference Singh2020).
To conclude, because we hypothesize that fictions are created mostly to attract attention, we do not hypothesize that there is any specific value in the information included in The Lord of the Rings or in Harry Potter. Imaginary worlds, we propose, are appealing because they meet the “input conditions” of our cognitive dispositions geared toward exploration (Sperber & Hirschfeld, Reference Sperber and Hirschfeld2004), just as romances and tragedies meet the input conditions of our preferences for love and social competition (Nettle, Reference Nettle2005a, Reference Nettle, Gottschall and Wilson2005b). Because fiction makers can intensify such attention-grabbing stimuli in the fictions, they grab our attention, even if the information is totally useless in real life.
Obviously, fictions tap into several kinds of human interests and our paper is about just one of them. Thus, our paper does not suggest that everybody should prefer fictions with imaginary worlds. A parallel can be made with cuisine: Sugar is clearly an important cultural attractor in the cultural evolution of recipes, but not all recipes include sugar and not everybody likes candies and pastries. In other words, although some people prefer consuming fictions about familiar places and comforting stories, others would preferably consume fictions with imaginary worlds and adventurous journeys. In the next section, we will argue that this variability in cultural preferences is evoked by the variability in the strength of exploratory preferences at the individual level.
4. The evolution and psychology of exploratory preferences
Our hypothesis is that the cultural preference for imaginary worlds relies on our exploratory preferences, driving our motivation to explore novel environments. As Tolkien put it himself, “part of the attraction of The Lord of the Rings,” and other fictions with imaginary worlds, relies on the “intrinsic feeling of reward” we experience when “viewing far off an unvisited island or the towers of a distant city” (letter to Colonel Worskett, September 20, 1963). This statement is very close to the one of Shigeru Miyamoto, the creator of Zelda, who reported that he “wanted to create a game world that conveyed the same feeling you get when you are exploring a new city for the first time” (1989). Such fictions would, thus, “tap into that deeply-seated human desire to travel, seek out new experiences, and absorb new knowledge about the world” (Etchells, Reference Etchells2019). In this section, we explain further the ultimate and proximate mechanisms behind exploratory preferences.
4.1. The evolution of exploratory preferences and capacities: the fitness benefits of exploration
From an evolutionary point of view, there is a broad consensus to say that exploring the environment is especially adaptive for mobile species, as it leads to discovering new vital resources such as food, finding mates and habitats, avoiding predators, and learning new action–outcome associations (Cashdan & Gaulin, Reference Cashdan and Gaulin2016; Chambon, Thero, Findling, & Koechlin, Reference Chambon, Thero, Findling and Koechlin2018; Gottlieb & Oudeyer, Reference Gottlieb and Oudeyer2018; Hayden & Niv, Reference Hayden and Niv2020; Hewlett, van de Koppel, & Cavalli-Sforza, Reference Hewlett, van de Koppel and Cavalli-Sforza1982; Hills, Reference Hills2006; MacDonald & Hewlett, Reference MacDonald and Hewlett1999; Miner, Gurven, Kaplan, & Gaulin, Reference Miner, Gurven, Kaplan and Gaulin2014; Panksepp, Reference Panksepp2005). For several species in several ecologies, such benefits outweigh to a certain point the costs of exploration (e.g., energetic loss, economic costs, risks of injury, and opportunity costs). Humans were particularly shaped by this selection pressure. Long-distance dispersal has been an important component of human migrations, allowing fast colonization of new territories all over the planet (Alves et al., Reference Alves, Arenas, Currat, Sramkova Hanulova, Sousa, Ray and Excoffier2016). Moreover, during the majority of its evolutionary history, Homo sapiens led a nomadic way of life (Lee, Reference Lee1966), similar to several other species who are known to travel across space (Chapman et al., Reference Chapman, Hulthén, Wellenreuther, Hansson, Nilsson, Brönmark, Hansson and Åkesson2014).
In line with the idea that there are fitness benefits derived from spatial exploration, studies have shown that nonhuman animals are endowed with specific capacities to explore their environment, recall the location of resources, determine the best navigation route between resources, and reorient when approaching locations from new perspectives (Rosati & Hare, Reference Rosati and Hare2012). Importantly, these capacities vary according to the ecology of the species (Healy, Dekort, & Clayton, Reference Healy, Dekort and Clayton2005; Platt & Brannon, Reference Platt and Brannon1996; Rosati, Rodriguez, & Hare, Reference Rosati, Rodriguez and Hare2014). For instance, chimpanzees exhibited more accurate spatial memory than bonobos across contexts, supporting predictions from these species' different feeding ecologies: wild chimpanzees depend more on patchily distributed fruit, whereas bonobos depend more on homogeneously distributed resources such as terrestrial herbaceous vegetation (Rosati & Hare, Reference Rosati and Hare2012). Fishes' spatial behavior too is very flexible, and this plasticity would rely on homologous cognitive mechanisms as those identified in mammals and birds (Broglio, Rodríguez, & Salas, Reference Broglio, Rodríguez and Salas2003). Importantly, these adaptations are not limited to purely cognitive capacities; they also extend to reward orientation. For instance, tamarins who feed on an ephemeral, dispersed food source (e.g., insects) and travel through large territories are much more likely to travel to a smaller, closer reward or a larger, more distant reward than marmosets who feed on a localized, immobile food (gum and sap exuding from trees) and, consequently, face little pressure to travel long distances for food (Stevens, Rosati, Ross, & Hauser, Reference Stevens, Rosati, Ross and Hauser2005).
Finally, it is important to note that such cognitive capacities are hypothesized to be the evolutionary precursor to goal-directed cognition: Several disciplines ranging from behavioral ecology to molecular genetics provide evidence that, for instance, problem-solving is a cognitive system born out ancient space-foraging behaviors (Hills, Reference Hills2006; Hills & Stroup, Reference Hills and Stroup2004; Hills, Todd, & Goldstone, Reference Hills, Todd and Goldstone2010). Such capacities are shared across all mobile organisms. Fishes' spatial behavior, for instance, is as elaborate as the ones of land vertebrates (Broglio et al., Reference Broglio, Rodríguez and Salas2003). Even further away from humans in the phylogenetic tree, bacteria's attempts to go back to resourceful environments through turns after removal from food is a strategic-foraging behavior (Korobkova, Emonet, Vilar, Shimizu, & Cluzel, Reference Korobkova, Emonet, Vilar, Shimizu and Cluzel2004; Neidhardt & Curtiss, Reference Neidhardt and Curtiss1996). In humans, cognitive maps are applied to non-spatial domains such as conceptual thinking (Behrens et al., Reference Behrens, Muller, Whittington, Mark, Baram, Stachenfeld and Kurth-Nelson2018; Epstein, Patai, Julian, & Spiers, Reference Epstein, Patai, Julian and Spiers2017; Jacobs, Reference Jacobs2003), and exploratory preferences are rallied to domain-general decision-making processes (Daw, O'Doherty, Dayan, Seymour, & Dolan, Reference Daw, O'Doherty, Dayan, Seymour and Dolan2006; Hills et al., Reference Hills, Todd and Goldstone2010; Le Heron et al., Reference Le Heron, Kolling, Plant, Kienast, Janska, Ang and Apps2019). Crucially, there is empirical evidence that preferences for spatial exploration in foraging tasks are correlated with preferences for cognitive exploration in problem-solving tasks (Hills & Stroup, Reference Hills and Stroup2004).
4.2 The behavioral manifestations of exploratory preferences: animal exploration, wayfinding, and environmental esthetics
Such preferences for spatial exploration are blatant when studying animal behavior. Several experimental studies revealed how curious about novel environments animals can be. In a famous study, rats were found to spend more time exploring novel environments than exploiting familiar ones (Berlyne, Reference Berlyne1950). Since then, the novelty-based theory of exploration and curiosity has been supported by several studies in the nonhuman animal literature, with various empirical tests, from the “open-field arena” test for mice (Berlyne, Reference Berlyne1970; Peeler & Nowakowski, Reference Peeler and Nowakowski1987) to the “visual novelty preference” test for monkeys (e.g., Fagan, Reference Fagan1970; Gunderson and Sackett, Reference Gunderson and Sackett1984). Rats can learn to find their way around a maze in the absence of rewards (Byrne, Reference Byrne2013; Reed & Adams, Reference Reed and Adams1996), suggesting that the opportunity to explore is intrinsically rewarding (Sabbatini et al., Reference Polizzi di Sorrentino, Sabbatini, Truppa, Bordonali, Taffoni, Formica, Baldassarre and Visalberghi2014). Rats and pigeons prefer multiple-choices paths over no-choice, shorter paths leading to the same reward (Bown, Read, & Summers, Reference Bown, Read and Summers2003; Catania, Reference Catania1980; Catania & Sagvolden, Reference Catania and Sagvolden1980; McDevitt, Pisklak, Spetch, & Dunn, Reference McDevitt, Pisklak, Spetch and Dunn2018). More recent studies with two-choice tests showed that animals are more interested by unfamiliar objects compared to familiar ones. At this stage, it has been proved to be the case for bottlenose dolphins, Pacific white-sided dolphins, beluga whales (Guarino, Yeater, Lacy, Dees, & Hill, Reference Guarino, Yeater, Lacy, Dees and Hill2017), rhesus macaques (Englerova, Klement, Frynta, Rokyta, & Nekovarova, Reference Englerova, Klement, Frynta, Rokyta and Nekovarova2019; Wang & Hayden, Reference Wang and Hayden2019), and orangutans (Borel, Ajzenherc, Moncel, Saint Jalme, & Krief, Reference Borel, Ajzenherc, Moncel, Saint Jalme and Krief2016). Also, it is worth noting that exploration is still preferred when the payoff is removed: In so-called non-instrumental tasks, animals observe novel stimuli even if they cannot act on them (Gottlieb & Oudeyer, Reference Gottlieb and Oudeyer2018). More surprisingly, exploration is attractive to the point that animals are willing to pay a supplementary cost to keep exploring (FitzGibbon, Lau, & Murayama, Reference FitzGibbon, Lau and Murayama2020; Hughes, Reference Hughes2007; Oudeyer, Gottlieb, & Lopes, Reference Oudeyer, Gottlieb and Lopes2016).
In humans, one obvious place to look for exploratory preferences is studies of wayfinding. Wayfinding is generally defined as the ability to move around efficiently and find the way from a starting point to a destination (Montello, Reference Montello, Shah and Miyake2005). It is different from pure locomotion. Locomotion depends on sensory-motor systems interacting with an immediate surrounding, whereas wayfinding invokes higher-level cognitive systems to maintain orientation relative to the distal environment. Experimental research has shown that performance in wayfinding is predicted by both cognitive capacities (e.g., visuospatial memory and mental rotation ability) and individual preferences. For instance, individuals who take pleasure in exploring places tend to have a good sense of direction (Muffato, Toffalini, Meneghetti, Carbone, & De Beni, Reference Muffato, Toffalini, Meneghetti, Carbone and De Beni2017) and perform better in spatial tasks (Carbone, Meneghetti, & Borella, Reference Carbone, Meneghetti and Borella2020; Muffato, Meneghetti, & De Beni, Reference Muffato, Meneghetti and De Beni2016, Reference Muffato, Toffalini, Meneghetti, Carbone and De Beni2017). In particular, Pazzaglia, Meneghetti, and Ronconi (Reference Pazzaglia, Meneghetti and Ronconi2018) showed that a significant part of the variability in the performance was explained by an aggregate measure of pleasure in exploring. Interestingly, the strength of the relationship between preferences and wayfinding tasks seems to depend on how difficult the task is: the tougher the task, the stronger the relationship (Pazzaglia et al., Reference Pazzaglia, Meneghetti and Ronconi2018; Weisberg, Schinazi, Newcombe, Shipley, & Epstein, Reference Weisberg, Schinazi, Newcombe, Shipley and Epstein2014).
Another area of research is known as “environmental esthetics,” a domain of empirical psychological research which investigates the elements of settings to which people are attracted the most (Balling & Falk, Reference Balling and Falk1982; Falk & Balling, Reference Falk and Balling2010; Herzog, Reference Herzog1984; Reference Herzog1985; Herzog & Bryce, Reference Herzog and Bryce2007; Herzog & Smith, Reference Herzog and Smith1988; Ikemi, Reference Ikemi2005; Kaplan, Reference Kaplan and Nasar1988; Kaplan & Kaplan, Reference Kaplan and Kaplan1989; Ruso, Renninger, & Atzwanger, Reference Ruso, Renninger, Atzwanger, Voland and Grammer2003; Ulrich, Reference Ulrich1979). Scholars working in environmental esthetics have conducted a great number of experimental studies to investigate the existence and the nature of universal preferences regarding environments. One of the key findings of this research program is that environments and landscapes are typically better rated and thus preferred when the settings signal an opportunity to gather information through exploration (e.g., the picture shows a trail that disappeared around a corner). Therefore, what makes an environmental setting appealing is the promise of further novel information, causally inferred from cues indicating that an enrichment in knowledge is a possibility. It is important to note here that these preferences are said to be automatic, unconscious, and intuitive. More often than not, participants were unable to explain their choices when rating the landscapes (Kaplan, Reference Kaplan1987), sustaining the hypothesis that sometimes “preferences need no inferences” nor explicit judgments (Zajonc, Reference Zajonc1980).
4.3. The cognitive and neural mechanisms underpinning exploratory preferences: plasticity, the exploitation–exploration trade-off, and the dopamine system
The study of wayfinding abilities and environmental preferences demonstrates the existence of specific preferences for spatial exploration. Exploratory preferences are part of a broader set of personality traits related to the meta-trait “plasticity.” The trait plasticity reflects the degree to which an organism is prone “to generating new goals, new interpretations of the present state, and new strategies to pursue existing goals” (DeYoung, Reference DeYoung2013, Reference DeYoung, Mikulincer, Shaver, Cooper and Larsen2015). Behavioral plasticity and exploratory preferences are inseparable notions because spatial exploration requires to be able to flexibly adapt to changing environments, otherwise exploration is too costly (Rojas-Ferrer, Thompson, & Morand-Ferron, Reference Rojas-Ferrer, Thompson and Morand-Ferron2020; Sol, Sayol, Ducatez, & Lefebvre, Reference Sol, Sayol, Ducatez and Lefebvre2016). From an empirical perspective, exploratory preferences are best studied through two important constructs that are highly correlated (DeYoung, Reference DeYoung, Mikulincer, Shaver, Cooper and Larsen2015; George & Zhou, Reference George and Zhou2001; Gocłowska, Ritter, Elliot, & Baas, Reference Gocłowska, Ritter, Elliot and Baas2019; Gottlieb, Oudeyer, Lopes, & Baranes, Reference Gottlieb, Oudeyer, Lopes and Baranes2013; Li et al., Reference Li, Li, Huang, Kong, Yang, Wei and Liu2015; McCrae, Reference McCrae1993): (1) the Big Five trait Openness to experience in personality psychology (e.g., Carbone, Meneghetti, and Borella, Reference Carbone, Meneghetti and Borella2019, Reference Carbone, Meneghetti and Borella2020; Meneghetti, Grimaldi, Nucci, & Pazzaglia, Reference Meneghetti, Grimaldi, Nucci and Pazzaglia2020; Pazzaglia et al., Reference Pazzaglia, Meneghetti and Ronconi2018) and (2) novelty-seeking in the neuroscience of decision-making (e.g., Costa, Tran, Turchi, & Averbeck, Reference Costa, Tran, Turchi and Averbeck2014; Krebs, Schott, Schütze, & Düzel, Reference Krebs, Schott, Schütze and Düzel2009). It is worth noting that wayfinding inclinations have been shown to be positively associated with Openness to experience (Carbone et al., Reference Carbone, Meneghetti and Borella2020; Meneghetti et al., Reference Meneghetti, Grimaldi, Nucci and Pazzaglia2020).
To further understand how exploratory preferences work at the proximate level, researchers designed the “bandit” task and its variants (Schulz & Gershman, Reference Schulz and Gershman2019): In its most basic design, the learner must choose between pulling a lever with known but degressive reward (i.e., to exploit) or a lever with unknown payoff, which is the exploratory choice (Cohen, McClure, & Yu, Reference Cohen, McClure and Yu2007; Daw et al., Reference Daw, O'Doherty, Dayan, Seymour and Dolan2006; Gershman, Reference Gershman2018; Le Heron et al., Reference Le Heron, Kolling, Plant, Kienast, Janska, Ang and Apps2019). It has been found that humans use a combination of both directed and random exploration strategies, with novelty cues as informative “bonuses” (Chakroun, Mathar, Wiehler, Ganzer, & Peters, Reference Chakroun, Mathar, Wiehler, Ganzer and Peters2020; Gershman, Reference Gershman2018; Gottlieb & Oudeyer, Reference Gottlieb and Oudeyer2018; Schulz & Gershman, Reference Schulz and Gershman2019; Wilson, Geana, White, Ludvig, & Cohen, Reference Wilson, Geana, White, Ludvig and Cohen2014). The exploitation–exploration trade-off is a classic problem in reinforcement learning. It corresponds to an evolutionary dilemma all mobile organisms face: They constantly need to arbitrate between exploiting a well-known (but maybe decreasing) source of resources or explore to find unknown (but maybe better) opportunities (Mehlhorn et al., Reference Mehlhorn, Newell, Todd, Lee, Morgan, Braithwaite and Gonzalez2015). The computational theory of reinforcement learning claims that the high-level goal of any learning agent is to obtain as much reward as possible, even if it is delayed (Dubey & Griffiths, Reference Dubey and Griffiths2020; Gozli, Reference Gozli2018) and it supports the idea from behavioral ecology that knowledge acquisition, prompted by novelty-based exploration, aims at optimizing future rewards (Brändle, Wu, & Schulz, Reference Brändle, Wu and Schulz2020; Dubey & Griffiths, Reference Dubey and Griffiths2020; Oudeyer et al., Reference Oudeyer, Gottlieb and Lopes2016).
The study of animal exploration in the previous sub-section suggested that rewards associated with exploratory behavior are different from and independent of external rewards present in the environment. Thus, across species, it is generally accepted that there is an intrinsic motivation to explore novel environments (Gottlieb et al., Reference Gottlieb, Oudeyer, Lopes and Baranes2013; Gottlieb & Oudeyer, Reference Gottlieb and Oudeyer2018; Liquin & Lombrozo, Reference Liquin and Lombrozo2020a, Reference Liquin and Lombrozo2020b). What are the neural bases for this intrinsic motivation to explore? There is persistent evidence that the dopamine system, known to be at the basis of rewards across several species (e.g., Baumann, Dames, Kühnel, and Walz, Reference Baumann, Dames, Kühnel and Walz2002; Hills, Reference Hills2004; Schultz, Reference Schultz1998, Reference Schultz2015), reacts specifically to novel stimuli which do not involve any primary reward (Düzel, Bunzeck, Guitart-Masip, & Düzel, Reference Düzel, Bunzeck, Guitart-Masip and Düzel2010; Horvitz, Stewart, & Jacobs, Reference Horvitz, Stewart and Jacobs1997; Kakade & Dayan, Reference Kakade and Dayan2002; Reed, Mitchell, & Nokes, Reference Reed, Mitchell and Nokes1996). Data from experiments with injections of a selective dopamine transporter inhibitor show that dopamine crucially enhances novelty-related value (Costa et al., Reference Costa, Tran, Turchi and Averbeck2014). A pivotal event-related functional magnetic resonance imaging (fMRI) study has demonstrated that novel pictures activated the mid-brain substantia nigra and ventral tegmental area (SN/TVA) more than rare, arousing and behaviorally relevant pictures (Bunzeck & Düzel, Reference Bunzeck and Düzel2006). That is, in the absence of reward, the dopamine system is activated by novel stimuli rather than interesting but more familiar ones. This finding supports the idea that “novelty can serve as its own reward” (Knutson & Cooper, Reference Knutson and Cooper2006) and is very much in line with the novelty-based theory of exploration (Bromberg-Martin, Matsumoto, & Hikosaka, Reference Bromberg-Martin, Matsumoto and Hikosaka2010; Gottlieb & Oudeyer, Reference Gottlieb and Oudeyer2018; Kidd & Hayden, Reference Kidd and Hayden2015). Interestingly, the reaction of the dopamine system to novel stimuli has been interestingly referred to as “novelty bonuses” (Frank, Doll, Oas-Terpstra, & Moreno, Reference Frank, Doll, Oas-Terpstra and Moreno2009; Kakade & Dayan, Reference Kakade and Dayan2002; Koster, Seow, Dolan, & Düzel, Reference Koster, Seow, Dolan and Düzel2016; Krebs et al., Reference Krebs, Schott, Schütze and Düzel2009; Krueger, Wilson, & Cohen, Reference Krueger, Wilson and Cohen2017; Sutton, Reference Sutton1990). Finally, empirical results from fMRI studies also show that, although the neuronal system coding for novel information-seeking behavior recruits the dopamine-based reward system, non-explorative choices (i.e., exploiting existing information) recruit different brain regions (Blanchard & Gershman, Reference Blanchard and Gershman2018; Chakroun et al., Reference Chakroun, Mathar, Wiehler, Ganzer and Peters2020; Costa, Mitz, & Averbeck, Reference Costa, Mitz and Averbeck2019; Daw et al., Reference Daw, O'Doherty, Dayan, Seymour and Dolan2006). Exploration, therefore, accounts for a specific neuronal and cognitive domain (Blanchard & Gershman, Reference Blanchard and Gershman2018).
4.4. The variability of exploratory preferences: life stage and ecological conditions
Finally, we go back to the ultimate level, to explain why and how such exploratory preferences are flexible, and vary according to the local environment and the life stage of the individual (Baumard, Reference Baumard2019; Frankenhuis, Panchanathan, & Nettle, Reference Frankenhuis, Panchanathan and Nettle2016; Jacquet, Safra, Wyart, Baumard, & Chevallier, Reference Jacquet, Safra, Wyart, Baumard and Chevallier2019; Nettle, Reference Nettle2019). Indeed, behavioral sciences have shown that organisms flexibly allocate resources such as energy and time to spatial exploration in a way that maximizes biological fitness (Charnov, Reference Charnov1976; Kaplan & Gangestad, Reference Kaplan and Gangestad2015; Stephens, Brown, & Ydenberg, Reference Stephens, Brown and Ydenberg2014). More specifically, all foraging species face an exploration–exploitation trade-off, that is, the dilemma between choosing either an exploitative or an exploratory option, be it for spatial foraging, choice making, or problem solving (Hills, Todd, Lazer, Redish, & Couzin, Reference Hills, Todd, Lazer, Redish and Couzin2015; Mehlhorn et al., Reference Mehlhorn, Newell, Todd, Lee, Morgan, Braithwaite and Gonzalez2015). Therefore, the strength of exploratory preferences should vary according to ecological conditions (because the costs and benefits of exploration depend on the resources of the environment) and the life stage (because the costs and benefits of exploration vary with one's life stage).
4.4.1. Exploratory preferences and life stages
In several animals, individuals' life stage impacts their exploratory strategy: Typically, they go through an early period of exploration followed by a later period of exploitation (Cohen et al., Reference Cohen, McClure and Yu2007; Morgan, Suchow, & Griffiths, Reference Morgan, Suchow and Griffiths2020; Stansfield & Kirstein, Reference Stansfield and Kirstein2006). This is the case because exploration is most adaptive when the individual knows little about the world, and juveniles from all species have less knowledge compared to adults (Blanco & Sloutsky, Reference Blanco and Sloutsky2020, Reference Blanco and Sloutsky2021). But juveniles from different species don't explore at the same rate or to the same extent. We argue that species with parental care explore more because the major costs associated with exploration (e.g., resource shortage risk) are outweighed by parental caregiving investments. Humans are a case in point here because human's juvenile period is longer than in any other species and allows for a long early protected period which can be devoted to cognitive and spatial exploration (Del Giudice, Reference Del Giudice2014; Kaplan & Gangestad, Reference Kaplan and Gangestad2015). In fact, studies have shown that although chimpanzees start being autonomous around 3 years old, individuals in modern hunter–gatherers societies are still dependent on their parents and kin up until 25 years old for feeding and protection (Bogin, Reference Bogin1997; Kaplan, Hill, Lancaster, & Hurtado, Reference Kaplan, Hill, Lancaster and Hurtado2000). This suggests that, although primates need to stop exploring very early on, human children can continue to play, learn, and explore their environment for a very long period of time before reaching puberty (Del Giudice, Reference Del Giudice2014). Researchers agree that human's prolonged childhood is central to our unique intelligence (e.g., Gopnik et al., Reference Gopnik, O'Grady, Lucas, Griffiths, Wente, Bridgers and Dahl2017; Piantadosi and Kidd, Reference Piantadosi and Kidd2016; Tomasello, Reference Tomasello2019).
Several evolutionary anthropologists and developmental biologists argue that this extended childhood stage is, in fact, a pivotal human adaptation. This life stage would have evolved in humans so that juveniles could have had the opportunity to acquire new foraging skills (Kaplan et al., Reference Kaplan, Hill, Lancaster and Hurtado2000; Kaplan & Robson, Reference Kaplan and Robson2002), social skills (Flinn & Ward, Reference Flinn, Ward, Ellis and Bjorklund2005), spatial skills (Piccardi, Leonzi, D'Amico, Marano, & Guariglia, Reference Piccardi, Leonzi, D'Amico, Marano and Guariglia2014), and reasoning skills (Buchsbaum, Bridgers, Skolnick Weisberg, & Gopnik, Reference Buchsbaum, Bridgers, Skolnick Weisberg and Gopnik2012). Therefore, skill learning would be the primary function of this long-lasting life stage. It can only work because, in return, older individuals compensate for the low productivity of juveniles with huge investments of time and resources (Kaplan et al., Reference Kaplan, Hill, Lancaster and Hurtado2000). It can be seen as an adaptive feedback loop or as an adaptive developmental division of labor (Buchsbaum et al., Reference Buchsbaum, Bridgers, Skolnick Weisberg and Gopnik2012; Gopnik, Reference Gopnik2020; Gopnik et al., Reference Gopnik, O'Grady, Lucas, Griffiths, Wente, Bridgers and Dahl2017; Kaplan et al., Reference Kaplan, Hill, Lancaster and Hurtado2000; Sumner, Steyvers, & Sarnecka, Reference Sumner, Steyvers and Sarnecka2019b). Such learning advantages of the extending childhood life stage explain why it evolved despite the heavy evolutionary costs associated with late reproduction.
In behavioral and cognitive sciences, much experimental evidence supports this idea. First, there is evidence that parental investments make early exploration possible: Cues of parental support or even the mere presence of a parental figure enhance exploratory behavior in children (Belsky, Goode, & Most, Reference Belsky, Goode and Most1980; Rubenstein, Reference Rubenstein1967; Snell-Rood & Snell-Rood, Reference Snell-Rood and Snell-Rood2020; Tottenham, Shapiro, Flannery, Caldera, & Sullivan, Reference Tottenham, Shapiro, Flannery, Caldera and Sullivan2019). Second, in some circumstances, children are more motivated to explore or better skilled at exploration compared to adults. From early in development, children seek causal explanations by asking questions about their environments (Liquin & Lombrozo, Reference Liquin and Lombrozo2020a, Reference Liquin and Lombrozo2020b). Younger children explore alternative uses of a tool more than older children (Defeyter & German, Reference Defeyter and German2003) and explore more flexibly alternative hypothesis compared to older children and even adults in problem-solving tasks (Gopnik et al., Reference Gopnik, O'Grady, Lucas, Griffiths, Wente, Bridgers and Dahl2017). Finally, children spend significantly longer time exploring their environments and explore at a higher rate than adults do (Blanco & Sloutsky, Reference Blanco and Sloutsky2019, Reference Blanco and Sloutsky2021; Gopnik et al., Reference Gopnik, O'Grady, Lucas, Griffiths, Wente, Bridgers and Dahl2017; Schulz, Wu, Ruggeri, & Meder, Reference Schulz, Wu, Ruggeri and Meder2019; Sumner et al., Reference Sumner, Li, Perfors, Hayes, Navarro and Sarnecka2019a, Reference Sumner, Steyvers and Sarnecka2019b).
To take one example, in an experiment with two sources of reward, children of age 5–12 collected fewer rewards compared to adults because they explored more (i.e., they switched more between the two sources of reward, even though one had higher payoffs), but, on the contrary, they were significantly more likely to detect an important change in reward opportunities, that a majority of adults missed because of their exploitative strategy (Sumner et al., Reference Sumner, Li, Perfors, Hayes, Navarro and Sarnecka2019a; on the costs of selective attention, see Blanco & Sloutsky, Reference Blanco and Sloutsky2019; on learning traps, see Rich & Gureckis, Reference Rich and Gureckis2018). The same kind of experimental design and the same results have been found with children doing a bandit task (Sumner et al., Reference Sumner, Steyvers and Sarnecka2019b), a reinforcement learning task (Liquin & Gopnik, Reference Liquin and Gopnik2019), and a change-detection task (Plebanek & Sloutsky, Reference Plebanek and Sloutsky2017): Children outperform adults because the latter missed information that children got through prolonged exploration. These results suggest that adults maximize payoffs at the cost of exploration and that, conversely, children invest more in exploration and thus miss fewer learning opportunities (Blanco & Sloutsky, Reference Blanco and Sloutsky2019).
Finally, further experimental research showed that children learn more from exploration compared to adults (Bonawitz, van Schijndel, Friel, & Schulz, Reference Bonawitz, van Schijndel, Friel and Schulz2012; Sim & Xu, Reference Sim and Xu2017a). For instance, experimental research in cognitive developmental psychology showed that children learn new, unexpected, or unusual causal relationships better (e.g., more rapidly, with fewer events) than adults do (Gopnik, Griffiths, & Lucas, Reference Gopnik, Griffiths and Lucas2015, Reference Gopnik, O'Grady, Lucas, Griffiths, Wente, Bridgers and Dahl2017; Lucas, Bridgers, Griffiths, & Gopnik, Reference Lucas, Bridgers, Griffiths and Gopnik2014; Sim & Xu, Reference Sim and Xu2017b).
4.4.2. Exploratory preferences and ecological conditions
Adaptive plasticity is the idea that individuals can adaptively express a range of different phenotypes depending on the state of the local ecology (Baumard, Reference Baumard2017; Frankenhuis & Nettle, Reference Frankenhuis and Nettle2020). More specifically, natural selection has favored psychological preferences that can flexibly adapt to different environments, so that the behavioral “programs” associated with such preferences maximize fitness in each of them. We argue that a substantial part of the variability in exploratory preferences, across time and populations, can be explained with adaptive plasticity. If exploration is more beneficial and less costly in some environments than in others, natural selection should have favored the expression of stronger exploratory preferences in such environments.
This is what behavioral ecologists commonly observe. Across species, exploration is sensitive to the level and steadiness of resources in the local environment (English, Fawcett, Higginson, Trimmer, & Uller, Reference English, Fawcett, Higginson, Trimmer and Uller2016; Humphreys et al., Reference Humphreys, Lee, Telzer, Gabard-Durnam, Goff, Flannery and Tottenham2015). For instance, studies with rats showed that exploratory behavior decreases with adversity in life (Spivey, Barrett, Padilla, & Gonzalez-Lima, Reference Spivey, Barrett, Padilla and Gonzalez-Lima2008). The same results have been found in other species such as black-capped chickadees (Rojas-Ferrer et al., Reference Rojas-Ferrer, Thompson and Morand-Ferron2020), 61 different parrot species (Mettke-Hofmann, Winkler, & Leisler, Reference Mettke-Hofmann, Winkler and Leisler2002), vampire bats (Carter, Forss, Page, & Ratcliffe, Reference Carter, Forss, Page and Ratcliffe2018), honeybees (Katz & Naug, Reference Katz and Naug2015), wild-spotted hyenas (Benson-Amram & Holekamp, Reference Benson-Amram and Holekamp2012), and orangutans (Damerius, Graber, Willems, & van Schaik, Reference Damerius, Graber, Willems and van Schaik2017; van Schaik et al., Reference van Schaik, Burkart, Damerius, Forss, Koops, van Noordwijk and Schuppli2016). This is also true in humans, who innovate and are more creative and opened to new experiences in affluent and safe societies (Baumard, Reference Baumard2019; Inglehart, Reference Inglehart2018). Empirical studies on human behavior have consistently supported this hypothesis: Individuals with high and steady levels of resources are more ready to explore novel information and new rewards (Frankenhuis et al., Reference Frankenhuis, Panchanathan and Nettle2016; Jacquet et al., Reference Jacquet, Safra, Wyart, Baumard and Chevallier2019; Nettle, Reference Nettle2019). In sum, several species including humans become more exploratory under the condition of relative safety.
These findings are best explained by the level of risks raised by exploratory behavior in different ecologies. In unsafe and poor ecologies, exploration is very risky, notably because if exploration doesn't pay off, one is left with nothing. Relatedly, the opportunity costs of exploration are higher in scarcity because one is better off providing for more pressing needs. Conversely, in more affluent and safer ecologies, such risks are lower: When surrounded by more resources, individuals can afford to lose some of them in the short term (Baumard, Reference Baumard2019). For instance, field observations suggest that wild orangutans avoid novelty whereas zoo orangutans are a lot more curious and explorative, with the very same tests. This contrast is best explained by the ecological differences between the two environments: Captive apes are fed and protected, and the risks of exploration such as resource shortage or predation are removed (Damerius et al., Reference Damerius, Graber, Willems and van Schaik2017; van Schaik et al., Reference van Schaik, Burkart, Damerius, Forss, Koops, van Noordwijk and Schuppli2016). In line with this idea, there is much evidence that animals adaptively reduce their exploration rate as predation risk increases (Verdolin, Reference Verdolin2006).
The example of the risk of resource shortage is another good test case here. Spatial exploration for foraging involves several risks related to resource collection. More specifically, it involves crucial opportunity costs such as waiting costs, that is, the costs associated with delayed (as opposed to immediate) collection of resources (Boon-Falleur, Baumard, & André, Reference Boon-Falleur, Baumard and André2020; Mell, Baumard, & André, Reference Mell, Baumard and André2021). Multiple optimal foraging models emphasize the discrepancy between the immediate risks of an exploratory strategy (e.g., the decrease of the resource levels during search time) and the positive value of the acquired information for future exploitation (e.g., Eliassen, Jørgensen, Mangel, & Giske, Reference Eliassen, Jørgensen, Mangel and Giske2007; Maspons, Molowny-Horas, & Sol, Reference Maspons, Molowny-Horas and Sol2019). In fact, exploration is best seen as an investment which is costly in the short term but beneficial in the long term, that is, a risky investment that organisms should only “prefer” to make in safe and affluent ecologies.
This line of argument suggests that it is only in an affluent environment that humans should afford to invest more in unpredictable exploratory activities (Baumard, Reference Baumard2019). It is also true if we consider the cultural evolution of human societies. It is well-established that, during the twentieth century, economic development is associated with more tolerance, more optimism, more interest in science, and less interest in religion (Inglehart, Reference Inglehart2018; Norris & Inglehart, Reference Norris and Inglehart2004). Similar observations can be made over the longer term: Economic development in ancient societies is associated with more tolerance (Martins & Baumard, Reference Martins and Baumard2020; Safra, Chevallier, Grèzes, & Baumard, Reference Safra, Chevallier, Grèzes and Baumard2020) and more exploration (Baumard, Reference Baumard2019; de Courson & Baumard, Reference de Courson and Baumard2019). To sum up, when resources are high, exploration is less risky and, thus, more likely to be advantageous. Therefore, under such conditions, phenotypic plasticity adaptively promotes and enhances exploratory preferences.
5. Exploratory preferences explain the cultural distribution of imaginary worlds
Although preferences for exploration are a human universal, the strength of exploratory preferences varies greatly from one individual to another. In this section, we argue that the variability in exploratory preferences partly explain individual differences in the preference for fictions with imaginary worlds, and therefore the cultural distribution of imaginary worlds across time, space, and population. We derive three predictions from this idea: (1) fictions with imaginary worlds should be more attractive to people high in Openness to experience, a personality trait measure used as a proxy for exploratory preferences, (2) younger individuals, for which exploration is less costly and more advantageous, should be more drawn to imaginary worlds compared to older individuals, and (3) individuals living in more affluent environments, where exploration is less risky and more adaptive, should have higher preferences for imaginary worlds.
5.1 Imaginary worlds should be more attractive to people higher in Openness to experience
If imaginary worlds co-opt our exploratory preferences, the appeal for imaginary worlds should be associated with Openness to experience, a component of the “Big Five” related to exploratory preferences (Carbone et al., Reference Carbone, Meneghetti and Borella2020; Meneghetti et al., Reference Meneghetti, Grimaldi, Nucci and Pazzaglia2020). This is indeed the case. In a recent paper, Nave, Rentfrow, and Bhatia (Reference Nave, Rentfrow and Bhatia2020) studied the association between personality traits and the “liking” of movies in Facebook users (N = 3.5 million). Using the same dataset, we show that higher scores in Openness to experience are associated with a preference for imaginary worlds (Dubourg et al., Reference Dubourg, Thouzeau, de Dampierre and Baumard2021). It is worth noting that, by contrast, fictions with imaginary worlds are associated with lower levels of extraversion, conscientiousness, agreeableness, and neuroticism (Dubourg et al., Reference Dubourg, Thouzeau, de Dampierre and Baumard2021).
Because Openness to experience is positively correlated with intellectual curiosity and higher academic achievement (Hakimi, Hejazi, & Lavasani, Reference Hakimi, Hejazi and Lavasani2011; Sorić, Penezić, & Burić, Reference Sorić, Penezić and Burić2017), it can be further predicted that people with a preference for fictions with imaginary worlds should have higher academic achievement. This is also the case. Consumers of fictions with imaginary worlds seem to be highly educated, compared to the general population. For instance, 82.4% of the survey with science fiction and fantasy fans report being educated to the university level or above (Menadue & Jacups, Reference Menadue and Jacups2018). This compares to 46% of the United States population, 50% of Australians, and 46% of the United Kingdom population (Organization for Economic Co-Operation and Development, 2017). This reinforces previous audience data that found “astonishing” high levels of education among the science fiction readership (Berger, Reference Berger1977, p. 236). Future research could notably test whether people who preferably consume fictions with imaginary worlds also preferably read informative non-fiction books, as well as other world-dominant fictions such as historical novels.
Openness to experience is also associated with more exploration in the social domain. People high in Openness to experience are indeed more likely to be tolerant of diversity, liberal, opened to new lifestyles, and opposed to right-wing political orientations (Butler, Reference Butler2000; Sibley & Duckitt, Reference Sibley and Duckitt2008). We, thus, predict that people who like imaginary worlds should be, overall, more politically tolerant, and socially liberal. This reasoning also predicts that people who enjoy imaginary worlds should prefer fictions exploring new social roles (Mar, Reference Mar2018; Mar & Oatley, Reference Mar and Oatley2008). In line with such predictions, recent research has observed that fans of fictions with imaginary worlds are more politically progressive and committed against prejudice (Besson, Reference Besson2021). Future empirical research could test this prediction.
In sum, we hypothesize that the preferences for imaginary worlds should be associated with a cluster of cultural preferences (e.g., more progressive political opinions and higher consumption of informative essays and historical novels) associated with Openness to experience.
5.2. Imaginary worlds should be more attractive to children, teenagers, and young adults
In section 4.4.1, we reviewed research showing that children have stronger exploratory preferences compared to adults and, under some conditions, even explore more than adults. For instance, in multiple experiments, they explored longer and generalized information from fewer events. We showed that this extended life stage can, in fact, be seen an adaptation: Major costs associated with exploration are outweighed by parental investments, so that children can “afford” to be more explorative. We, thus, hypothesize that children are more attracted to fictions with imaginary worlds.
In line with this hypothesis, we commonly observe that humans develop an early interest for imaginary worlds. Psychological research has shown that, very early on, children produce imaginary worlds in their heads (Silvey & MacKeith, Reference Silvey and MacKeith1988; Taylor, Mottweiler, Aguiar, Naylor, & Levernier, Reference Taylor, Mottweiler, Aguiar, Naylor and Levernier2020). More importantly, consumers of fictions with imaginary worlds are typically (and, according to our hypothesis, accurately) stereotyped as young (Besson, Reference Besson2015; Jenkins, Reference Jenkins1998; Proctor & McCulloch, Reference Proctor and McCulloch2016). For instance, young readers are targeted by massively consumed novels in the fantasy genre, sometimes adapted for the screen with unparalleled successes. Let's think of Tolkien's The Hobbit (1937), Horowitz's Groosham Grange (1988), Riordan's Percy Jackson (2005–2009), Rowling's Harry Potter (1997–2007), Pullman's His Dark Materials (1995–2000), Saint-Exupéry's The Little Prince (1943), and Colin's Hunger Games (2008–2020). All these highly successful books with their famous imaginary worlds are edited in the children's collections from the publishing houses. We can further illustrate this point by mentioning the Walt Disney Studios, the single most productive and lucrative studios for children films. It is commonly observed that a majority of Disney films is based on the exploration of imaginary worlds (Elza, Reference Elza2014). For instance, in Alice in Wonderland (1951), Alice decides to explore Wonderland (as in the children novel by Lewis Carroll) and in Peter Pan (1953), Wendy and his two brothers decide to explore the imaginary world of NeverLand.
Empirical research partly confirmed this prediction. In a study on the correlation between literary taste patterns and social differentiation in Finland, age was the only variable which significatively decreased the liking of speculative fictions (a proxy for fictions with imaginary worlds; see sect. 2), whereas age had no such effect on other fictional genres (Purhonen, Gronow, & Rahkonen, Reference Purhonen, Gronow and Rahkonen2009). In Dubourg et al. (Reference Dubourg, Thouzeau, de Dampierre and Baumard2021), we found a significative and negative correlation between age and a preference for movies with imaginary worlds: such movies tend to be liked by younger people. To our knowledge, our approach is the only one that consistently explains this strong association between the preference for imaginary worlds and the age of consumers.
5.3. Imaginary worlds should be more attractive to people living in more affluent environments
In section 4.4.2, we have also shown that exploration is more adaptive in predictable and affluent ecologies, because of the risk variable involved in the exploration–exploitation trade-off. If the level of resources in the local environment is high and steady enough, individuals can afford to delay potential benefits and to take risks: they become more motivated to explore the real world. We, therefore, predict that, both at the level of the individual and at the level of societies, affluence is a good predictor of the preference for imaginary worlds.
This prediction is in line with empirical findings. A recent empirical survey (N = 909) provided insights about the socioeconomic status of science fiction and fantasy fans (Menadue & Jacups, Reference Menadue and Jacups2018). To the question about the income satisfaction level, most respondents answered: “I do well enough” (54.1%), and 32% answered “I'm happy with what I have” or “I have more than I need.” Such readers also have curious and open-minded psychological traits. For instance, 95.2% reported they found new and unfamiliar ideas easy to understand. Because this study is based on a selected sample, we lack data to compare these results with the socioeconomic status of readers of other fictional genres, but it still confirms that consumers of fictions with imaginary worlds fit this general prediction. In another study, in Finland, speculative fictions were found to be significantly more read by people with higher income, whereas income as a variable had no such effect on the consumption of other genres (Purhonen et al., Reference Purhonen, Gronow and Rahkonen2009).
At the global level, our hypothesis predicts that imaginary worlds should be more popular and, therefore, emerge in economically more developed countries. In line with this idea, the very first imaginary world comparable in size to the real world, Tolkien's world Arda, is extremely recent. Before 1914 (the first developments of Arda by Tolkien), humans had long begun to produce and consume literary fictions, and they didn't lack any cognitive abilities that would have prevented them to invent large imaginary worlds with much background information. However, only a few stories had developed large imaginary worlds (e.g., Dante's Hell) and virtually none had been precisely described and mapped. We argue that the late appearance of imaginary worlds is explained by the evolution of the strength of exploratory preferences. For a long time, people's exploratory preferences were too weak to give rise to the production of imaginary worlds in fictions. Economic development made such preferences adaptive in some populations, and only then could imaginary worlds appear and be culturally successful.
In fact, modern imaginary worlds first appeared in the United Kingdom (Wolf, Reference Wolf2012), which was at the time the leading country in terms of GDP per capita (Bolt & van Zenden, Reference Bolt and van Zanden2020), and then mostly developed in the Euro-American sphere (e.g., France, the United States, and Germany) and in the 1950s in Japan. By contrast, although Jules Verne was first translated in Chinese in the early twentieth century and inspired Chinese writers to write science fiction and fantasy stories during the late Qing dynasty and early Republican era, fictions based on imaginary worlds remained marginal in the Chinese literature during the twentieth century (Jiang, Reference Jiang2013). Imaginary worlds started to become popular first in Hong-Kong and Taiwan, which started to develop in the 1970s, and really became mainstream in mainland China in the turn of the new millennium, that is, 20 years after the take-off of the Chinese economy (Song, Reference Song2013).
In a recent empirical study (Dubourg et al., Reference Dubourg, Thouzeau, de Dampierre and Baumard2021), we studied the evolution of the share of fictions with imaginary worlds in 11 countries, since the beginning of the nineteenth century, with data extracted from Wikidata (N = 44,608). In most countries, when GDP per capita increases, the share of fictions with imaginary worlds rises too. We also studied another indicator of success, namely the box-office of films with imaginary worlds in the United States (Dubourg et al., Reference Dubourg, Thouzeau, de Dampierre and Baumard2021). An indicator of the success of films with imaginary worlds (IWS indicator) was computed by subtracting, for a given year, the mean box-office of films with imaginary worlds and the mean box-office of films with no imaginary worlds, in the United States. This indicator is highly and positively correlated with the GDP per capita in the United States. It even becomes positive: Above a certain threshold of GDP, films with imaginary worlds generate more revenues than others do. It would be interesting to test whether the quantitative analysis of the success of films with imaginary worlds and its association with affluence indicators replicate in non-Western developed countries (e.g., Japan and Korea) and developing countries (e.g., India and Nigeria).
Finally, it is interesting to note that as societies become safer and more affluent, people seem to grow out of the fondness for imaginary worlds at a later and later age. We argue that this is the case because, in such local environments, people should afford to remain explorative longer. This would explain why a new target audience has recently emerged, the “young adults,” with associated editorial collections (i.e., YA literature) often specialized in speculative fictions with imaginary worlds. Future empirical research could focus on this prediction that in more economically developed societies, across both time and space, the mean age of fans of imaginary worlds is higher. This hypothesis lays the ground for a more general research program in behavioral sciences on the longer hold of children cultural preferences in modern societies.
6. Exploratory preferences shape the content and form of fictions with imaginary worlds
We argued that human's exploratory preferences, determined by ecological conditions and the life stage of the individuals, explain the cultural distribution of imaginary worlds, that is, the individual differences in the preference for imaginary worlds, the timing of their appearance in cultural history, and the variability of their success across societies and across populations. In this section, we focus on modern and contemporary culture. We hypothesize that, as soon as fictions with imaginary worlds emerge as a competitive market, their form and content should be shaped by what best co-opt humans' exploratory preferences. The basic idea is that cultural items compete for the attention of audiences and, therefore, producers are likely to intensify appealing stimuli to increase the success of their works. For instance, Walt Disney's Mickey co-opts our visual preference for baby faces. It has been shown that the evolution of its design is driven by this preference: Across the last few decades, Mickey progressively became cuter, that is, more baby-like, with larger heads and more doting eyes (Gould, Reference Gould2008; Hinde & Barden, Reference Hinde and Barden1985). Similarly, because films have competed for the attention of moviegoers since the beginning of cinema, they have undergone continual changes. For example, over time, films have gotten faster (shot lengths have decreased) and darker (luminance have decreased), to better grab the attention of the viewers and improve their engagement in the film (Cutting, Brunick, DeLong, Iricinschi, & Candan, Reference Cutting, Brunick, DeLong, Iricinschi and Candan2011).
This “superstimulus” hypothesis (or “stimulus intensification” hypothesis) posits that, as soon as enough people were safe and rich enough that strong exploratory preferences emerged and made imaginary worlds culturally successful (i.e., after Second World War in Europe and North America), producers started to invent, selectively retain, and cumulatively refine features that best exploited exploratory preferences, to make their imaginary worlds more appealing than other ones. From this point, several predictions follow, two of which we detail in this section: across time, (1) information background should increase in fictions with imaginary worlds, and (2) more particularly, fictions with imaginary worlds should generate more and more “paratexts” (i.e., information devices that surround the fiction; Genette, Reference Genette1997).
6.1. Fictions with imaginary worlds should generate more and more non-narrative background
At the ultimate level, the function of exploration is to accumulate new information and maximize the usefulness of knowledge for future rewards (see sect. 4.1). This means that any new information about the real world (e.g., the localization of a foraging site) and any device making such information easier to learn (e.g., navigation systems) should attract the human mind. If the appeal for imaginary worlds indeed exploits such exploratory preferences, fiction makers should target this evolutionary function by generating, in their fictions, more and more apparently useful background information leading to a better grasp of the imaginary environment.
This prediction appears to be validated by the cultural evolution of imaginary worlds. First, as time goes by, imaginary worlds are more and more precisely detailed in literary texts (Wolf, Reference Wolf2021). Tolkien's world is a case in point. It is remarkable that never before in the history of literature had there been such a comprehensive imaginary world. Imaginary worlds existed in ancient literature (see Table 2), but they were never thoroughly described and documented. Since Tolkien, although, several imaginary worlds have been extensively developed with much information about the settings (Wolf, Reference Wolf2012). This observation is perfectly in line with our prediction: Now that humans' exploratory preferences are heightened, any piece of background information about the world becomes an even more interesting stimulus that fiction makers can target. Why would consumers memorize so well so much information that yet only applies to the imaginary world? For instance, fans of Harry Potter know Quidditch rules, fans of Star Wars know the names of the planets, fans of Game of Thrones know the geography of Westeros, and fans of Pokémon know the evolution of each specimen (e.g., Delle, Reference Delle and Brenner2015). Such a list of useless (but effective-seeming) pieces of information about imaginary worlds that hundreds of millions of people learn, retain, and debate about, could go on and on (Besson, Reference Besson2015). A case in point: such information is not only memorized, but also organized and stored online by fans: There is a “fandom” encyclopedia-like website for each famous imaginary world. To take one salient example, the online encyclopedia about Star Wars had 167,792 pages at the time of writing this paper.
Because background information has become an attention-grabbing stimulus, it should be intensified in fictions, that is, it should eventually become a “superstimulus.” This is what fiction makers do: They expand the amount of information made available for a given imaginary world. Wolf (Reference Wolf2017b) defines the “size” of an imaginary world as “the number of world data describing it.” Importantly, this should not be mixed up with the “scope” of an imaginary world, which is the extent of the space covered by the imaginary world (e.g., an imaginary village, an imaginary planet, etc.). An imaginary world can be large but poorly described and, conversely, small in scope but very dense in details. We can now refine our prediction by saying that, because of humans' exploratory preferences, imaginary worlds with more world data should be more successful at a given time and, therefore, world data should increase. In sum, the size of imaginary worlds should progressively be intensified, regardless of their scope. Therefore, our hypothesis posits that exploratory preferences explain why “we are drawn to master what can be known about a world which always expands beyond our grasp” (Jenkins, Reference Jenkins2006).
This is a strategy largely observed in video games, which rapidly evolved to include open imaginary worlds that the players can freely explore with “sandbox” gameplay: The player is given a great degree of freedom in the gradual discovery of the world. A case in point: In most open-world video games, the map of the imaginary worlds is not revealed right away, and one of the player's goal is to unveil it (Bartle, Reference Bartle2004). This is also the case in films and novels. Imaginary worlds are never precisely described at once, for instance, at the beginning of the fiction. Rather, information about the other-worldly settings comes progressively as the narrative unfolds, keeping our curiosity alive. For instance, the Star Wars galaxy is composed of several environments that are revealed in the course of the story. Crucially, such planets, to which characters travel by high-speed spaceships, are highly different from one another (e.g., Tatooine is a desert planet, Dagobah is a jungle planet, and Hoth is an ice planet). Another option for producers of fictions is to add more world data in other fictions set in the same imaginary world (Besson, Reference Besson2015; Wolf, Reference Wolf2021), and sometimes from different media platforms. This gives rise to transmedial imaginary worlds (Konzack, Reference Konzack and Wolf2018; Rebora, Reference Rebora2016) and to media franchises (Besson, Reference Besson2015). For instance, Rebora (Reference Rebora2016) argued that fantasy is “the best fitting literary ground for any transmedial expansion.” Let's note that among the 20 highest-grossing media franchises, more than half are fictions set in an imaginary world (Wikipedia, 2021).
To further test this prediction that background information about imaginary worlds increases and makes fictions with imaginary worlds more successful, one could further operationalize the quantity of world data about imaginary worlds and look at the evolution of this measure over time. To do this, one could use a semantic tool that encodes spatial structure of worlds from literary texts (Louwerse & Zwaan, Reference Louwerse and Zwaan2009). It has already been done with Tolkien's Middle Earth (Louwerse & Benesh, Reference Louwerse and Benesh2012). If applied to enough literary fictions, this tool could measure the cultural evolution of the informational complexity of imaginary worlds.
6.2. Fictions with imaginary worlds should generate more and more paratexts
Information about imaginary worlds can be more or less organized. It can be transmitted in a natural way, as in everyday life, but it can also use artificial devices such as maps, lists, or genealogies that greatly increase our ability to manipulate, store, and organize vast amounts of information (Goody, Reference Goody1986). We, thus, predict that information devices leading to a better grasp of the information embedded in fictions with imaginary worlds should be appealing and should, therefore, increase in number across recent time.
In line with this prediction, paratexts (Genette, Reference Genette1997) such as maps, guidebooks, appendices, lists, family trees, footnotes, or glossaries recently emerged in fictions with imaginary worlds, and became rapidly mainstream (Saler, Reference Saler2012). In particular, maps attract our attention because they deliver spatial information about imaginary worlds. The fictional map in Stevenson's Treasure Island (1881–1882), one of the first imaginary maps, is partly what drew consumers to this book (Wolf, Reference Wolf2012). This suggests that people at that time rapidly became curious with non-narrative fictional artifacts. Since then, a substantial and growing part of fantasy fictions is released with maps of the imaginary worlds (Ekman, Reference Ekman2013). Some maps, such as Thror's map in The Hobbit and the Marauder's map in Harry Potter, are used by the characters in the fictions. Even if a map is not provided by the producers of the fiction, some fans always compile spatial information in the fiction, create their own maps and put them online. This suggests that virtually all imaginary worlds have been mapped, be it by fiction makers or by fans. Maps are even more useful in open-world video games, because gamers interact with the game world (Haggard & Chambon, Reference Haggard and Chambon2012; Nguyen, Reference Nguyen2019; Tanenbaum & Tanenbaum, Reference Tanenbaum and Tanenbaum2009). Such video games always include an interactive map which allows the players to find their way around the world (Akchelov & Galanina, Reference Akchelov and Galanina2016; Nitsche, Reference Nitsche2008; Wolf & Perron, Reference Wolf and Perron2014).
But maps are not the only kind of information device made available by fiction makers to increase the appeal of their fictions. Fictions with imaginary worlds can also include informational texts, either in the fictions (e.g., the Encyclopedia Galactica in Foundations) or published after the release of the fiction as real-world books (e.g., Pandorapedia, James Cameron's encyclopedia of Avatar's imaginary world Pandora, and Harry Potter Schoolbooks). Such guidebooks can even be written by other people than the creator and still be successful (e.g., Philip Pullman's His Dark Materials: The Definitive Guide by Laurie Frost). They almost all contain, among other things, hierarchized information about creatures, locations, and plants, as well as family trees of the protagonists. This information is also widely available online, notably in the “fandom” encyclopedias. “Pottermore” is yet another illustration of this trend: Rowling herself created a website for the sole purpose of providing more background information about the “Wizarding World” of Harry Potter. Finally, guidebooks for video games are highly successful worldwide. For instance, the guidebook to Final Fantasy VIII is an actual best-seller, with 2.2 million sold copies over the world. It is part of the Japanese guide series Square Enix companion books, which is by far the best-selling guide series about imaginary worlds, and largely outgrows any series of encyclopedia of the real world in terms of revenue.
6.3. Limitations of the “superstimulus” hypothesis
Further research should investigate if this stimulus intensification has limits: Not enough world data could be disappointing or even boring, but, conversely, too much world data could be bewildering, frustrating, or too complex, exactly like the attraction to novelty (Andersen et al., Reference Andersen, Schjoedt, Price, Rosas, Scrivner and Clasen2020; Clark, Reference Clark2018; Kidd, Piantadosi, & Aslin, Reference Kidd, Piantadosi and Aslin2012; Kiverstein, Miller, & Rietveld, Reference Kiverstein, Miller and Rietveld2019). This limitation has been mentioned in other studies about stimuli intensification in fictions. For instance, Gessey-Jones et al. (Reference Gessey-Jones, Connaughton, Dunbar, Kenna, MacCarron, O'Conchobhair and Yose2021) studied the network of character interactions in George R. R. Martin's epic novels, A Song of Ice and Fire and found that the degrees of the most connected characters reflect a cognitive limit on the number of concurrent social connections that humans tend to maintain (see also Dunbar, Reference Dunbar, Kenna, MacCarron and MacCarron2017). It is likely that similar limits constrain the size of imaginary worlds.
The example of No Man's Sky (2016) is compelling: Through procedural generation of worlds, this video game includes over 18 quintillion planets, with as many different imaginary environments which can be explored with complete autonomy from the players. Before its release, it was expected to be a major hit in the video game industry precisely because it was announced to be the biggest explorable game world (Morris & Hartas, Reference Morris and Hartas2004). Yet it has been less successful compared to other exploration-based video games (only 68% of users' evaluations are positive on the popular ranking website for video games Steam). Further experimental research could assess whether people are reluctant about virtually infinite imaginary worlds and, if so, why. This is a burning issue for the video game industry because technological improvements make such developments technically feasible, whereas our cognitive constraints might not make them desirable.
7. Discussion
7.1. Remaining questions and alternative explanations
In research fields interested in fictions, there has long been a focus on “who” and “how” questions, about plots and protagonists, at the expense of “where” questions, about settings, probably because of our narrative-oriented understanding of fictions. As Ryan recently wrote, “narrative space remains a relatively unexplored territory” (Ryan, Reference Ryan2014). In this paper, we have provided evidence that narrative spaces, and in particular imaginary worlds, are central in modern fictions partly because they tap into human's preferences for exploration, which have been co-opted by cultural evolution for entertainment. Obviously, more research is needed to further test this theory. For instance, we need to be able to quantify the size of imaginary worlds, that is, the amount of background information associated with a particular world. The existence of clusters of cultural preferences (e.g., imaginary worlds and historical novels) should also be tested rigorously. Besides, several questions remain unsolved: What exactly is the cultural advantage of fictions with imaginary worlds over non-fictions describing the real world (e.g., history books and travel books)? Why are medieval fantasy and space opera so attractive, compared to other imaginary worlds? Is there an ideal cognitive trade-off between too much imagination and too much similarity with the real world? To our knowledge, no empirical research has tested the limits of the processing of novelty in human culture.
On another note, other sets of cognitive mechanisms might play an important role in the appeal for imaginary worlds, and we do not rule out other complementary explanations. For example, systemizing seems central in the appreciation of world-dominant fictions, be it with an imaginary world (e.g., Star Wars), or not (e.g., Sherlock Holmes). Indeed, although humans are lured by new environments, there is much evidence that they also prefer rich and organized ones (Kaplan, Reference Kaplan1987). This might come from our drive to systemize (Baron-Cohen, Reference Baron-Cohen2002) and from our cognitive mechanisms which make us intuitively think about plants and animals in highly structured ways (Atran, Reference Atran1998). This hypothesis is consistent with recent cognitive frameworks stating that curiosity seeks both novelty and complexity to maximize knowledge acquisition (Brändle et al., Reference Brändle, Wu and Schulz2020; Dubey & Griffiths, Reference Dubey and Griffiths2020). This echoes the “encyclopedic impulse” scholars in cultural and literary studies targeted to explain the attractiveness of world-dominant fictions (Besson, Reference Besson2015; Eco, Reference Eco1997; Wolf, Reference Wolf2012). It leads to testable predictions about sex differences in cultural preferences for highly structured imaginary worlds.
7.2. Exploratory preferences and other cultural trends
Conversely, our hypothesis could explain other trends in the cultural evolution of fictions. For instance, fanfictions (i.e., fictional writings written by fans and based on previous canonical fictional works) have become highly mainstream in several countries (e.g., dōjinshi in Japan and Star Trek fanzine in the United States) and are beginning to be taken very seriously by consumers and the publishing industry. Gamers too started to computationally create more content for their favorite video games, with the intention of sharing it with others. Such alterations of the games are called “mods” (short for “modifications”). More and more game studios create mod tools to ease this process on the fan's side. This massive cultural phenomenon, with fans reshaping and improving video games for the sole benefit of the mod community, and free of charge, has been overlooked in psychological and cognitive research fields (Poor, Reference Poor2014; Sotamaa, Reference Sotamaa2010). It has never been put in parallel with literary fan fictions nor with the literature on exploration and curiosity. More generally, the motivation behind unpaid user-generated content (UGC) has mainly been explained with social benefits (Chavez et al., Reference Chavez, Ruiz, Curras and Hernandez2020; Crowston & Fagnot, Reference Crowston and Fagnot2018; Daugherty et al., Reference Daugherty, Eastin and Bright2008; Omar & Dequan, Reference Omar and Dequan2020; Sun et al., Reference Sun, Dong and McIntyre2017) and economic incentives (Poch & Martin, Reference Poch and Martin2015). We argue that our hypothesis could better explain this drive to create new cultural content without any direct return on investment. It relates to the examples we mentioned of several animals that explore even in the presence of other primary rewards, and even with experimentally added costs. Humans seek so much for new information that this may well push fans to create new content, even at some costs. In other words, according to our hypothesis, modders, writers of fan fictions and other participants of UGC, are best seen as curious explorers.
Our hypothesis could also contribute to clarify the cognitive bases of other types of fiction, such as interactive books, films, and TV series (e.g., Black Mirror: Bandersnatch, 2018), or, to a lesser extent, crime, mystery, horror, and detective fictions. These genres arguably tap into our exploratory preferences, but not into spatial exploration per se. Rather, they seem to exploit an uncertainty-based form of exploration. We, therefore, believe fictions in these genres target the related mechanisms designed to minimize uncertainty, seek reason-based explanations (Gottlieb et al., Reference Gottlieb, Oudeyer, Lopes and Baranes2013; Grodal, Reference Grodal2010; Liquin & Lombrozo, Reference Liquin and Lombrozo2020a), and detect and evaluate arguments (Mercier, Reference Mercier2016).
7.3. Imaginary worlds and the cultural evolution of fictionality
Our theory about the cultural evolution imaginary worlds can be put in the wider perspective of the cultural evolution of fiction. We argue that fictionalization has been a gradual process. As it has long been noted by literary historians and literary theorists (Bakhtin & Emerson, Reference Bakhtin and Emerson1984; Cave, Reference Cave1999; Lavocat, Reference Lavocat2016; Lévi, Reference Lévi1995; Lu, Reference Lu1994, Reference Lu2000; Paige, Reference Paige2011; Postel, Reference Postel2019), the most ancient fictions such as The Greek and Indian epics, the Greek and Latin tragedies, and the Arthurian romances, all tend to feature characters (e.g., gods, heroes, and kings) already known to the audience, with plots that are themselves known (e.g., the Trojan War and the story of Tristan and Iseult), and in already familiar worlds (e.g., Troy and the court of King Arthur). As Bakhtin famously wrote: In the epics, “the special interest in the ‘end’ (‘How will the war end?,’ ‘Who will win?,’ ‘What will happen to Achilles?,’ etc.) is totally excluded.” In ancient fictions, creators mostly invent new versions of old stories, with new scenes or secondary characters. The other possibility is that they take the form of anecdotes, legends, chuanqi (i.e., “tales of the strange” or “records of the anomalies”), novella (i.e., news), which they relate as true, as having really happened to someone they know. It is only gradually that the stories will become fictionalized both by the inclusion of completely new characters and intrigues, and by the progressive abandonment of the claim to veracity.
Thus, we argue that the important question is not whether people in ancient societies “believed” in their fictions (Veyne, Reference Veyne1988) but to which degree their narratives were truly fictional. We argue that in ancient time, such stories were not “as fictional” as they can be in modern societies. This would explain why, for instance, Chinese people recognized fictions not before the Ming and Ch'ing dynasties, only when they “dehistoricized” narratives (Lu, Reference Lu1994). Consistently, data-driven studies show that it is only in the seventeenth century in England that fiction makers started to write fictions about “nobodies,” that is, characters that are completely unknown to the audience (Paige, Reference Paige2020). It marked the beginning of the “novel” (Bakhtin & Emerson, Reference Bakhtin and Emerson1984; Cave, Reference Cave1999; Lu, Reference Lu1994; Paige, Reference Paige2011). In this perspective, imaginary worlds can be thought as the ultimate step of this fictionalization process that started centuries ago: After having fictionalized the events and the protagonists, fiction makers started to fictionalize the settings, giving themselves even more freedom to intensify all stimuli in the fictions.
7.4. Broader concluding remarks
We now discuss some broader conclusions our paper brings about. First, our paper adds further support to the cultural attraction theory according to which human culture is influenced by our cognitive biases (Boyer, Reference Boyer2018; Claidière et al., Reference Claidière, Scott-Phillips and Sperber2014; Claidière & Sperber, Reference Claidière and Sperber2007; Morin, Reference Morin2016; Scott-Phillips et al., Reference Scott-Phillips, Blancke and Heintz2018; Singh, Reference Singh2021). Culture is neither faithfully nor randomly transmitted, but rather reconstructed in a way shaped by our cognition. The cultural evolution of imaginary worlds is one example of this cultural evolutionary process: A set of cognitive mechanisms which evolved to solve the adaptive trade-off between exploration and exploitation drives the evolution of fictions with imaginary worlds. Following this view, we disentangled two main paths through which imaginary worlds culturally evolved: Imaginary worlds have emerged and changed over the course of history (1) because our exploratory preferences evolved to adapt to crucial ecological changes, through phenotypic plasticity (sect. 5), and (2) because producers of fictions target and exaggerate already preferred stimuli (sect. 6). Crucially, the combination of these two processes explains both the universality and the cultural variability of imaginary worlds.
Second, we expect that our findings could be relevant for literary theory, cultural history, and fiction study. For instance, the existence of distinct genres such as horror, comedy, and detective fictions has been said to derive from the involvement of distinct sets of cognitive mechanisms (Clasen, Reference Clasen2010; Clasen et al., Reference Clasen, Kjeldgaard-Christiansen and Johnson2018; Clasen & Platts, Reference Clasen, Platts, Vanderbeke and Cooke2019; Fishelov, Reference Fishelov1995; Grodal, Reference Grodal2010, Reference Grodal2017). Our study extends this on-going research program and could also support the cognitive studies of fictional media, such as cinema (Jullier, Reference Jullier2018; Tan, Reference Tan2018). We believe our hypothesis, if supported by more empirical evidence, can also be used by scholars in evolutionary psychology and computational history. The consumption of fictions with imaginary worlds could be used as a behavioral proxy to measure the evolution of exploratory preferences, offering insights into their adaptive flexibility to the changing environments.
Finally, it is our belief that this paper could be relevant outside the research domain, for education and fiction production. The fact that children are intrinsically captivated by imaginary worlds suggests that such fictions should be brought into the classrooms. The links between curiosity and learning being increasingly understood (Gordon et al., Reference Gordon, Breazeal and Engel2015; Wade & Kidd, Reference Wade and Kidd2019), cultural attraction theory could be a relevant and useful framework to design curiosity-based learning interventions. More generally, exploring which types of fictions exploit which cognitive mechanisms brings about new predictions about the socio-psychological determinants of the attractiveness of specific cultural items. Further research in this domain could lead to more fine-grained and evidence-based individual suggestions of which fictions to read or watch. This would have direct implications for the fiction industry and their recommendation algorithms (Nave et al., Reference Nave, Rentfrow and Bhatia2020).
Acknowledgment
The authors thank Sacha Altay, Mélusine Boon-Falleur, Valerian Chambon, Charles de Dampierre, Léo Fitouchi, Hugo Mercier, Olivier Morin, and Valentin Thouzeau.
Financial support
This work was supported by FrontCog funding (ANR-17-EURE-0017).
Conflict of interest
None.
Target article
Why imaginary worlds? The psychological foundations and cultural evolution of fictions with imaginary worlds
Related commentaries (32)
All non-real worlds provide exploration: Evidence from developmental psychology
Am I present in imaginary worlds? Intentions, actions, and flow in mediated experiences and fiction
Autism and the preference for imaginary worlds
Brave new world: Imaginative fictions offer simulated safety and actual benefits
Cognitive exploration drives engagement and re-engagement with imaginary worlds, but not spatial exploration as predicted by evolutionary theory
Determining the markers of a preference for imaginary worlds fiction calls for comparisons across kinds of fiction readers and forms of exploration
Does progressive aphantasia exist? The hypothetical role of aphantasia in the diagnosis of neurodegenerative diseases
Exploration and arrangement in physical and social worlds
How granular should our explanations of fantastic fiction be?
Imaginary worlds are awesome: Awe provides a key to understanding the individual and social functions of imaginary worlds
Imaginary worlds are attractive because they simulate multiple adaptive problems and encode real-world information
Imaginary worlds pervade forager oral tradition
Imaginative processes in children are not particularly imaginative
Imagining our moral values in the present and future
Influences on the development of imaginary worlds
It's not fiction if you believe it: How imaginary worlds are derived from imaginary realities
Middle-earth wasn't built in a day: How do we explain the costs of creating a world?
Not just a hijack: Imaginary worlds can enhance individual and group-level fitness
Simulation, stories, and fictional worlds
Socioecology and fiction
The allure of the unknown in a tamed, mapped, and homogenized world
The call of the final frontier?
The deep history of imaginary worlds
The development of the imagination and imaginary worlds
The evolution of imagination and the adaptive value of imaginary worlds
The familiar appeal of imaginary worlds
Using imaginary worlds for real social benefits
Why do people create imaginary worlds? The case of Fanfiction
Why frightening imaginary worlds? Morbid curiosity and the learning potential of horror
Why imaginary worlds? The role of self-exploration within online gaming worlds
Young children are not driven to explore imaginary worlds
“Never Land”: Where do imaginary worlds come from?
Author response
Imaginary worlds through the evolutionary lens: Ultimate functions, proximate mechanisms, cultural distribution