A common view in behavioral approaches to literature is that the capacity to tell stories is adaptive (Gottschall and Wilson, 2005; Carroll, 2012). Notably, it has been argued that consuming fictions leads to acquire fitness-related knowledge (Sugiyama, 2001; Smith et al., 2017; Schniter et al., 2018; Nakawake and Sato, 2019; Sugiyama, 2021b), self-regulate one’s emotional states (Schaeffer, 1999; Gottschall and Wilson, 2005; Martin et al., 2018), simulate fake scenarios to be better prepared to face the real world (Tooby and Cosmides, 2001; Sugiyama, 2005; Mar and Oatley, 2008; Bloom, 2010; Gottschall, 2012; Clasen, 2019; van Mulukom and Clasen, 2021), or attract sexual mates (Miller, 2001). Evolutionary speaking, these hypotheses would hold only if our ancestors had faced a specific adaptive challenge that the behavior of producing or consuming fictions would have specifically solved (Tooby and Cosmides, 1992). It does not appear to be the case: neither of these hypotheses identify an adaptive function that is specific to narrative fictions (Carroll, 2012).

Let us take first the evolutionary theories which have proposed that the function of fictional narratives is to transmit adaptive information, be it about foraging (Sugiyama, 2001, 2021a), animals (Nakawake and Sato, 2019) or cooperation (Coe et al., 2006; Smith et al., 2017). The point is that such claims are not specific to narrative fictions, as factual narratives can also (and is, we assume, even more efficient to and more used to) transmit such generalizable knowledge. That is, communication and social learning had solved the adaptive problem of information transmission in such a way that fiction does not appear to enhance (Dunbar, 2003; Boyd et al., 2011; Boyd, 2018). Importantly, if many factual pieces of information, or ‘teaching moments’, as Sugiyama (2021b) call them, are embedded in fictions, they are precisely features that would make us doubt of the fictional status of the overall product. To put it in another way, if individuals truly wanted to convey relevant and important information about the world, they would not use invented content or pragmatic signals of fictionality (letting the consumers understand that the text is partly composed of falsehoods, for example, by introducing the story with “Once upon a time”). In a nutshell, we contend that telling an openly non-fictional story is a much more efficient way to transmit non-fictional information.

Similarly, the simulation hypothesis cogently explains why humans have the capacity to imagine or simulate hypothetical scenarios: because it solves the adaptive challenge of forecasting problems and working out solutions without actual practice (Oatley, 1999; Harris, 2000; Tooby and Cosmides, 2001). Imagination and foresight might be evolutionary adaptations (Suddendorf and Corballis, 2007; Fuentes, 2020; van Mulukom, 2020), but imagination is not fiction. Military strategists, projects managers, or engineers all need to imagine several potential futures in order to find the best options. But this activity does not fall under the category of fiction. In fact, a science fiction writer is likely to trade the credibility of a simulation for its potential in terms of entertainment, through the invention and ostensive exaggeration of attention-getting situations. In science-fiction, nuclear wars, aggressive IA, and alien invasion are the rigueur, not in military’s strategies related to actual potential threats. In other words, if speculative imagination aims at forecasting potential events to be better prepared to real life, this capacity seems now (and since quite a long time) to be used to invent and share narrative fictions that do not directly aim at being better prepared to face potential real situations, because they are too far from real events (Morin et al., 2019). From A Brave New world to 1984 to The Neuromancer, the worlds of science fiction are most often very dark, not because they aim at forecasting the future, but because they aim at entertaining the readers.

Other approaches posit that fictions are adaptive because they would train or enhance our social skills (Zunshine, 2006; Mar and Oatley, 2008). However, human social cognitive capacities, such as Theory of Mind (Gerrans, 2002; Tsoukalas, 2018), and human behavioral preferences, such as morality (Sperber and Baumard, 2012; Baumard et al., 2013; Tomasello, 2015), have been selected by natural selection and do not need fiction consumption to fully develop. If it was adaptive for individuals to be more cooperative, more cooperative individuals, that is individuals genetically more motivated to cooperate, would be naturally selected without the need to go through the implementation of the cognitive capacities needed for the behaviors of producing fictions.

To take a similar case, no one argues that humans need the consumption of fiction to enhance their sexual and romantic interests or to motivate themselves to raise children, because it is clear that if there were an adaptive advantage to being more motivated to have sex or to care for one’s children, then individuals who are more motivated to have sex or to care for their children would be naturally selected (Cosmides and Tooby, 2013). As a matter of fact, it is interesting to note that pornography or romance are seldom view as adaptive, despite the fact that they share the same fictional nature as other more legitimate fictions (Salmon, 2012). Besides, in the empirical literature, the effects of narrative fictions on the consumers’ beliefs or behavior are overall small (Gentile et al., 2009; Mulligan and Habel, 2013; Vezzali et al., 2015; Borum Chattoo and Feldman, 2017; Mumper and Gerrig, 2017; Dodell-Feder and Tamir, 2018; Rathje et al., 2021). Importantly, they are also elicited by factual narratives (Barnes and Black, 2021). For instance, studies comparing people who consumed fictional movies and others who watched documentaries found no evidence of differential effects on prosociality (LaMarre and Landreville, 2009).

The same counterarguments hold for the hypothesis that narrative fictions have the adaptive function to regulate one’s emotional states. Emotional instability derives from hard-wired cognitive mechanisms that make people react to various situations in adaptive ways (Cosmides and Tooby, 2000; Nettle, 2012; Al-Shawaf et al., 2016). More precisely, emotions, such as fear, shame, guilt, gratitude, or pride, are cognitive programs whose specific function is to coordinate other mechanisms that should be efficiently coordinated facing a specific (adaptive) problem (Al-Shawaf et al., 2016). These behavioral programs are already fine-tuned to each situation, so that fictions would in fact be counterproductive in terms of biological fitness if they substantially impacted their regulation. However, we do not make the (absurd) claim that narrative fictions do not change the emotions of the consumers. We rather argue that they did not evolve to perform such a function. And, again, this idea is supported by the fact that the (minor) affective effect is not fiction-specific: it is elicited, for instance, by music (Mehr et al., 2020; Savage et al., 2020).

To conclude this section, we argue that the current adaptive hypotheses do not account for the fact that producers invent narratives and do not explain why consumers pay attention to narrative. The proposed evolutionary functions are not specific to fictional narratives. It is not clear why humans would need to evolve fictions to regulate their emotions, transmit information, or forecast the future because evolving fiction would not be the most straightforward way to do that. Also, adaptive hypotheses do not explain why such fictions should depart from realistic narratives. Because fictional narratives do exist in human cultures, there should be a specific advantage for narratives to be fictional.

The other hypothesis posits that narrative fictions are by-products, and therefore did not evolve through natural selection. Within this framework, it is argued that fictions co-opt pre-existing cognitive capacities and preferences that evolved in the human mind for no reasons related to fictions, and that this explanation is sufficient to explain the existence, universality, and pervasiveness of fictions in human cultures. A version of this hypothesis has been famously called the “cheesecake hypothesis” by Pinker (1997). Cheesecakes exploit the cognitive mechanisms designed to make humans detect and like the taste of glucose, at the proximate level. Those mechanisms have been selected by evolution because the ingestion of glucose enhanced fitness in the environments in which the human mind evolved, at the ultimate level (Ramirez, 1990). Therefore, the preference that makes humans like cheesecakes evolved long before cheesecakes appeared. Masks are another good example: because they display visual patterns that are close to real faces (e.g., two points at the top of a round shape, one point at the bottom of the same round shape; Farroni et al., 2005), they meet the input conditions of the face recognition mechanism that evolved to identify individuals and understand their emotions (Sperber and Hirschfeld, 2004). As a consequence, they artificially trigger people’s face recognition mechanism and automatically grab their attention. On top of that, by exaggerating facial traits and facial expressions (e.g., bigger eyes, more colorful faces, etc.), they produce new, original, and often more powerful emotions.

Likewise, fictions would be pleasurable and attention-grabbing for the human mind, at the proximate level, because they would co-opt a myriad of cognitive preferences, that evolved before symbolic culture even emerged. In line with the by-product hypothesis, many studies brought evidence that fictions do co-opt cognitive preferences that evolved before fictions even existed (Table 1).

The by-product hypothesis explain well why fictions are attention grabbing (they meet the input conditions of many preexisting cognitive mechanisms). However, it does not explain why (1) producers produce fictions (what is the fitness advantage of creating worlds, characters, and plots?) and (2) why consumers consume fictions (what is the fitness advantage of spending so much time to learn about worlds, characters, and plots that do not exist?; André et al., 2020).

To put it in other words, the by-product hypothesis makes a strong hypothesis about the nature of human cognition. It indeed assumes that the interest in fictions is essentially a mismatch and that humans are not able to understand that they are wasting their time with imaginary characters and imaginary worlds. This is very possible in theory. For instance, pure psychoactive drugs (e.g., heroin) that are administered directly in the blood are very novel. They thus bypass adaptive information-processing systems and induce positive emotions that give a false signal of a fitness benefit. This signal hijacks mechanisms of “liking” and “wanting,” and is inherently pathogenic (Nesse and Berridge, 1997). However, even in the case of drug, this assumption should be considered with caution. For instance, the consumption of drugs that are eaten or smoked may very well be adaptive, notably against parasites. In line with the adaptive framework, the consumption of these drugs is lower for individuals whose brain is not mature enough to tolerate neurotoxins, and therefore adaptively varies with age, sex, and condition (e.g., pregnancy; Hagen et al., 2013). To conclude, it is possible that fictions are just the result of a mismatch, and hijack evolutionary ancient mechanisms. However, this mismatch hypothesis should really be used in the last resort (as in the case of directly administered drugs) when all possible explanations have failed. We believe that this is not the case for fictions.

In the next section, we propose a middle-ground solution that explains why the existence of fiction is adaptively plausible, for both the producers and the consumers, and why the content of fiction is so well tuned to the human mind.

Literary theorists and historians have long noticed the cross-culturally recurrent and entertaining features of fictions (which have also been called “themes,” “tropes,” or “patterns”) such as adventures, conflicts, love stories, imaginary worlds, monsters, gossip, authority, success, and the search of social status (Kato and Saunders, 1985, p. 232; Pavel, 1986, pp. 147–148; Campbell, 1993; Schaeffer, 1999, p. 241; Huang, 2001, pp. 60–61; Hogan, 2003; Booker, 2004). Evolutionary critics in the humanities and evolutionary social scientists brought evidence that such universal fictional features are influenced by the evolutionary history of the human mind (Carroll, 1995; Gottschall, 2008; Fisher and Salmon, 2012; Saad, 2012; Grodal, 2017). More recently, as we have seen in section The By-product Hypothesis (and the Problem of Fitness Benefits), these cross-cultural features have been linked to specific cognitive preferences (Table 1). In all, there seems to be a large and interdisciplinary consensus to say that narrative fictions include attractive and entertaining features. The question therefore is: Why are such features attractive and entertaining to the human mind?

We contend that such pleasurable features of fictions are very close to what evolutionary biologists called superstimuli (Tinbergen, 1969; Barrett, 2010). Many studies show that some species, in the course of their evolutionary history, recycled pre-existing attractive traits for new evolutionary relevant functions such as attracting mates (Lorenz, 1966; Krebs and Dawkins, 1978; Basolo, 1990; Ryan et al., 1990). For instance, because the female frog Physalaemus pustulosus had developed preferences for lower-frequency chuck sounds, males evolved the ability to produce such sounds to tap into this sensory preference (Ryan et al., 1990).

In nonhuman animals, this recycling of preexisting preferences usually emerges through biological selection. In humans, it can emerge through cultural evolution: producers use their expertise to target and refine stimuli that are already appealing to consumers (Lightner et al., 2022), so as to fulfill fitness relevant goals (Singh, 2020). We will explain what these goals are in the next sub-section.

We therefore argue that content features in fictions are superstimuli: they are crafted to resemble stimuli that were already appealing to the human mind, because of the natural selection of attention-orienting cognitive mechanisms, and of the pleasure systems rewarding the behavior of paying attention to such stimuli. This is a form of what psychologists have called “content-based attraction,” when the attraction and prevalence of a cultural item is favored by its content (Sperber, 1996; Claidière and Sperber, 2007; Scott-Phillips et al., 2018).

A question follows: Why are such stimuli attention-grabbing in the first place (in the real world)? This is where we fall back on the by-product hypothesis: such preferences for some stimuli (e.g., social information) evolved because humans endowed with them survived and reproduced better in the ancestral environments when the human cognition evolved.

In evolutionary and cognitive approaches to fictional content, superstimuli have already been studied in fictional texts (Jobling, 2001; Nettle, 2005a,b; Singh, 2019), in movies (Cutting et al., 2011; Andrews, 2012; Clasen, 2012; Cutting, 2016, 2021; Sobchuk and Tinits, 2020), in video games (Jansz and Tanis, 2007; Mendenhall et al., 2010), in artistic representations (Verpooten and Nelissen, 2010, 2012), and in cross-media approaches to fiction (Grodal, 2010; Barrett, 2016; Dubourg and Baumard, 2021). Let us note that such fictional superstimuli can be narrative superstimuli (e.g., how Marcel in Search of Lost Time reaches prestige), visual superstimuli (e.g., the form of Mickey), auditory superstimuli (e.g., the terrifying sounds in horror films), and other sensory superstimuli (e.g., the sense of control in open-world video games or in virtual reality games). Producers of fictions use any means available to them to make the most attention-grabbing superstimuli and therefore the most entertaining fictions.

Of course, the pleasure-inducing effect elicited by superstimuli in fictions is also elicited by some other cultural behavior and products, such as sport and news (Barrett, 2010, 2016). This is because the fiction industry is not the only one to target entertainment. However, the presence of superstimuli successfully isolate fiction from non-fiction, because superstimuli are never included in non-fictional narratives: the obligation to (try to) stick to real facts prevent, to a large extent, producers of non-fictional narratives to invent and exaggerate any feature (or else their epistemic reputation might suffer, and the benefits of attracting other people’s attention would be overweighted by the reputational costs of having deceived their audience). We contend that such a distinction is intuitive to consumers: they will continue to consume and positively evaluate fictions that they take pleasure from, while they will either stop consuming or negatively evaluate fictions that deceive the expectation to be entertained. Conversely, when they consume non-fictional narratives, such as a philosophical treatise, a political essay, or an history documentary, their primary goal is to learn things, so that they will not stop consuming the non-fiction if they are not entertained, and they will not base their evaluation on this criterion.

Why would producing fictions be adaptive? With the entertainment hypothesis, this question is the same as the following one: Why would attracting the attention of other people by inventing entertaining cultural items should bring any fitness benefit? We propose that, because they are highly attractive and entertaining, fictions can be used to fulfill any evolutionary relevant goal that needs others’ attention to be caught, be it signaling one’s values to potential mates (Miller, 2001) or cooperative partners (Bourdieu, 2010; André et al., 2020; André and Baumard, 2020; Dubourg et al., 2021b; Lightner et al., 2022), transmitting knowledge (Schniter et al., 2018; Nakawake and Sato, 2019; Sugiyama, 2021b), communicating social norms (Mar and Oatley, 2008; Ferrara et al., 2019), or selling products (Saad and Gill, 2000; Saad, 2012).

Consistently, narrative fictions seem to have been used (1) as recruitment technologies: they allow the producers of fictions to attract and potentially cooperate with individuals that matter to them, by signaling one’s qualities (e.g., their competence, their moral sense, and their intelligence) and therefore enhancing one’s reputation as a cooperative partner (Sperber and Baumard, 2012). For instance, in many countries at most time in history, cultural institutions and organizations aimed at spotlighting the producers of fictions, from the poetry contests (uta-awase) in Japan from the Heian period to the modern Nobel Prize in Literature and movie Academy Awards. Narrative fictions are also obviously used to (2) derive economic or material gains. This is clearly pictured in the form fiction production and fiction consumption took in large-scale societies, that of a massive (and highly lucrative) contract-based market.

Crucially, such adaptive goals need not be conscious or deliberate. They need not be the only motivations either: drawing on adaptive hypotheses that we reviewed in section State of the Current Hypotheses, producers of fictions can have other goals, such as transmitting knowledge (Sugiyama, 2021a). The association between both motivations of educating and entertaining people has produced a new form of cultural devices called “Edutainment” (Singhal, 2004; Anikina and Yakimenko, 2015), which we argue has emerged far back in human cultural history, embedding not only recent fictions (e.g., Dora the Explorer), but also ancient folktales (Sugiyama, 2021b) and other literary forms such as pre-17^th century European fairy tales.

According to this hypothesis, narrative fictions are sustained because they confer fitness benefits to the consumers too. First, let us note that the opportunity costs of fiction consumption seem rather low because people do not seem to consume fictions at the expense of other more “evolutionary relevant” activities such as sleeping, eating, and parenting. On the other hand, consumers can use fictions they liked to signal their skills (Veblen, 1899; Bourdieu, 1979; Lizardo, 2006, 2013). They can also use more culturally successful fictions they liked to signal their personality traits (Dubourg et al., 2021a), or to share cultural focal points for social coordination (Dubourg et al., 2021b,c). Besides, human minds have evolved specialized cognitive mechanisms to detect and use social markers for coordination (Nettle and Dunbar, 1997; Boyer, 2018). We propose that preferences for fictions have become relatively important markers in the ecology of modern cultural diversity, because of their signaling potential.

In all, we propose that humans did not specifically evolve the capacity to tell fictional stories, but they rather produce fictions thanks to a range of other adaptations (e.g., language, the capacity to simulate, Theory of Mind, and communicative inferences; Zunshine, 2006; Mellmann, 2012; Wilson, 2018). Yet, we do not consider fictions as “by-products,” because they clearly confer fitness benefits to the producers (André et al., 2020). We argue that fictions are “entertainment technologies” (Dubourg and Baumard, 2021): they are crafted by storytellers to artificially attract the attention of other people and then fulfill evolutionary-relevant goals (Singh, 2020). Obviously, fictions are not the only example of entertainment technologies. Sport, TV shows (Barrett, 2010, 2016), music (Dubourg et al., 2021a), and performing arts (Verpooten and Nelissen, 2010, 2012) are also entertainment technologies in the sense that they are created to trigger people’s attention, and are consumed because they exaggerate the features of phenomena (e.g., human voice and interindividual competition) that humans evolved to be interested in.

The entertainment hypothesis posits that people’s preferences are factors of attraction and thus drive the cultural evolution of fictions, because the producers of fiction’s goal is to make entertaining cultural products. Therefore, because people’s preferences vary, we expect narrative fictions to vary accordingly. More precisely, our framework predicts that the variability of preferences, which is explained and predicted by evolutionary psychologists and behavioral ecologists, impact the variability of cultural consumption. Here, we identify three main sources of the interdividual variability of evolved preferences: the life stage, the sex, and the conditions of the local ecology of the individuals. We propose that such causal factors of the variability of biological preferences can account for the cultural distribution of fictions across time and populations.

Not only are narrative fictions filled with appealing stimuli, but we also posit that, over time, such stimuli are selectively retained and cumulatively refined to better attract the attention of the consumers. The basic idea is that narrative fictions compete for the attention of consumers in what one could call an ‘entertainment economy’. Producers are therefore likely to intensify already appealing stimuli, to increase the success of their narrative fictions. Importantly, producers need not know the evolutionary origins of such and such preferences shaping the content of their creations. The selection and refinement of features at each generation and the ‘trial and error’ process are sufficient to explain the improvement of fictional features across time.

Some empirical findings suggest that this is the case. The most grounded example is undoubtedly the one of the ‘baby schema’ in visual fictions. The ethologist Konrad Lorenz hypothesized that a set of infantile features, such as a round face and big eyes, is perceived as ‘cute’, at the proximate level, because, at the ultimate level, having one’s attention caught by cute babies motivated parental caregiving investments and was therefore adaptive. Lorenz (1943) provided correlational evidence that this was the case. More recently, experimental research reached the same conclusion: pictures of babies that were parametrically manipulated to produce an enhanced baby schema (e.g., with rounder faces and larger eyes than real babies) were rated as cuter and as motivating more caregiving than photographs of babies that were both manipulated to produce low baby schema (e.g., with less round faces and smaller eyes) and photographs not manipulated at all (Glocker et al., 2009a). Using the same pictures (as experimental stimuli) and functional magnetic resonance imagining, another study showed that a specific brain system (the mesocorticolimbic system) is responsible for the emotional and behavioral response to cute babies (Glocker et al., 2009b). This line of research provide straightforward predictions, when applying the cumulative cultural evolution framework to the entertainment hypothesis: if producers of fictions select, refine and exaggerate appealing stimuli (to better tap into evolved preferences and make more entertaining fictions), protagonists should become cuter and cuter. This has been empirically shown with Walt Disney’s Mickey: the evolution of its design is driven by this preference for cute baby faces: across the last decades, Mickey progressively became cuter, that is, more baby-like, with larger heads and more doting eyes (Hinde and Barden, 1985; Gould, 2008).

Recent empirical work started to unveil other cumulative processes in the refinement of entertaining features in fictions. Godzilla grabs our attention because its height and strength would make it a very dangerous predator if it were real, and a quick look at its successive representations shows that it gets bigger and taller over time (Dominy, 2019; Sobchuk, 2019). More universally, movies grab our attention in part because the rapidity of the sequence of shots (i.e., the shot lengths) is suited to make the eye movements reevaluate each visual depiction: with 75 years of Hollywood film, Cutting et al. (2011) provided empirical evidence that shot lengths have significantly decreased, to enhance this control over the audience’s eye movements. In another work, we argued that imaginary worlds cumulatively evolved too, by including more and more information background and information devices that modern consumers find attractive (Dubourg and Baumard, 2021). The examples taken here all show cumulative processes: producers at each generation selectively retain and cumulatively refine fictional elements that seem to best fit their goal of entertaining their audiences. We suspect that many more superstimuli in fictions have been cumulatively refine in recent times, because of the tremendous growth of both fiction production and consumption allowing faster cumulative processes. Much more empirical research is needed to assess the way each superstimulus in narrative fictions is cumulatively selected and refined over cultural history.