Menninghaus et al. suggest that activation of the concepts <art> and <fiction> causes emotional distancing, a hypothesis that is both widely accepted in aesthetic philosophy (Kantian “disinterestedness”; see Stolnitz [Reference Stolnitz1961]) and experimentally demonstrable (Gerger et al. Reference Gerger, Leder and Kremer2014; Mocaiber et al. Reference Mocaiber, Pereira, Erthal, Machado-Pinheiro, David, Cagy, Volchan and de Oliveira2010; Reference Mocaiber, Perakakis, Pereira, Pinheiro, Volchan, de Oliveira and Vila2011a; Reference Mocaiber, Sanchez, Pereira, Erthal, Joffily, Araujo, Volchan and de Oliveira2011b; Van Dongen et al. Reference Van Dongen, Van Strien and Dijkstra2016; Wagner et al. Reference Wagner, Menninghaus, Hanich and Jacobsen2014). But their claim that <fiction> can be “activated far beyond the confines of art” (sect. 3) implies either that the concept <art> has meaningful boundaries for its possessor or the word “art” picks out a circumscribed set of objects. Both possibilities enjoy scant theoretical or empirical support. Normative art definitions face numerous difficulties (Davies Reference Davies2006), and so far no evidence suggests that lay or expert intuitions converge on one or another (Kamber Reference Kamber2011). Our lab recently extended Malt's (Reference Malt1990) study on concept beliefs and found categories like “art” and “music” treated quite differently from other artifacts or nominal and natural kinds: their boundaries were considered the least strict, yet determined by experts to a degree on par only with biological groupings. Descriptive art definitions, on the other hand, seem reasonable but rule out very little. Most features on Dutton's (Reference Dutton2009) list, for instance, arguably apply to nearly any complex human endeavor. We might shift the locus of explanation from ill-defined objects to some special mental states associated with them, but plausible models of aesthetic experience face a similar predicament. Any reasoned judgment about an object or event likely involves sensation, perception, categorization, contextual knowledge, and affect (Chatterjee & Vartanian Reference Chatterjee and Vartanian2014; Leder & Nadal Reference Leder and Nadal2014) or these elements plus causal reasoning and considerations of object history (Bullot & Reber Reference Bullot and Reber2013b; notably, Pearce et al. Reference Pearce, Zaidel, Vartanian, Skov, Leder, Chatterjee and Nadal2016, add disinterestedness to their list). This conceptual obscurity is not a problem for the authors' model but a considerable problem for aesthetic psychology, as the latter's objects of study are correspondingly vague. A possible solution expands the notion of fiction to encompass any signal believed to be neither true nor false.

In ethology, a signal is an action or property of an animal that guides another animal's behavior (Maynard-Smith & Harper Reference Maynard-Smith and Harper2003). An index is a signal that cannot be faked, for example, intensification of display colors caused by healthy diets (Blount et al. Reference Blount, Metcalfe, Birkhead and Surai2003). Call these signals_T because they are always true (or reliable, if you prefer). Other signals can be true or false (signals_TF); a bird may issue a call that normally indicates predators whether or not a predator is present (Munn Reference Munn1986). Although it is controversial whether nonhuman animals knowingly deceive, a signal's recipient must nonetheless judge its veracity before responding and can use contextual cues to do so (Heinen & Stephens Reference Heinen and Stephens2016). Yet humans seem (uniquely?) able to classify actions in a third way, as signals with no determinate truth value (signals_NDT). When someone yells “fire!” in a crowded theater, others must choose whether to flee or stay – unless the signaler is onstage in costume, in which case they understand that the action is not a claim about some occurrent state of the world. Invoking art or fiction through explicit labeling or contextual cues allows subsequent sensory inputs to be treated as signals_NDT. For organisms with the metacognitive capacity to represent beliefs and their relationships to reality, classifying input this way supports counterfactual thought (“suppose things were as she said”), thereby avoiding unnecessary action (“flee!”) or erosion of trust (“she lied”).

If correct, this view constrains investigation without species-wide art concepts or special aesthetic experiences. The objects of study are – minimally – things thought to be signals_NDT. Although Menninghaus et al. call fiction a subgroup of public representations, fiction in this expanded sense is a superordinate category that includes and thus links the disparate artforms studied by aesthetic psychologists. Art and fiction are effectively co-extensive.

Tooby and Cosmides (Reference Tooby and Cosmides2001; Cosmides & Tooby Reference Cosmides, Tooby and Sperber2000) have advanced a similar proposal in some detail, but restricted their discussion to public representations. Yet nonrepresentational art becomes especially intriguing on this view. There is obvious utility in our appetite for, and emotional responses to, fictional narratives, which illustrate social invariants from a safe remove (Mar & Oatley Reference Mar and Oatley2008). But it is unclear why instrumental music or abstract art elicit any emotions when they have no narratives and represent nothing – yet they do (Gabrielsson & Wik Reference Gabrielsson and Wik2003; Pelowski Reference Pelowski2015).

I have invoked ethology, but the capacity to classify input as signals_NDT need not be innate; it could be acquired across the life course, for instance, through pretend play (Weisberg Reference Weisberg2015). One might worry that people initially treat all events as signals_T (Gilbert Reference Gilbert1991), so the classification is simply illusory. But evidence runs contrary. Reduced amplitudes of an event-related potential associated with highly emotional stimuli have been observed for disturbing images presented as movie stills or artworks as opposed to documentary photographs (Mocaiber et al. Reference Mocaiber, Pereira, Erthal, Machado-Pinheiro, David, Cagy, Volchan and de Oliveira2010; Van Dongen et al. Reference Van Dongen, Van Strien and Dijkstra2016). Comparable emotional attenuation has been captured with facial electromyography, functional imaging, and heart rate monitoring (Gerger et al. Reference Gerger, Leder and Kremer2014; Mocaiber et al. Reference Mocaiber, Perakakis, Pereira, Pinheiro, Volchan, de Oliveira and Vila2011a; Reference Mocaiber, Sanchez, Pereira, Erthal, Joffily, Araujo, Volchan and de Oliveira2011b). A salient signal treated as true should not elicit attenuated emotional responses so quickly and consistently.

Of course, false signals_TF would predict the same pattern. But people dislike deception. If artworks were considered mere lies, our enthusiasm for them should have long ago waned (although it remains possible that they are classified as “white lie” signals_TF [Iñiguez et al. Reference Iñiguez, Govezensky, Dunbar, Kaski and Barrio2014]). Moreover, narratives explicitly labeled fictions nonetheless alter beliefs about the world (Green & Brock Reference Green and Brock2000; Marsh & Fazio Reference Marsh and Fazio2006; Strange & Leung Reference Strange and Leung1999). That signals_NDT seemingly cause both emotional attenuation and belief updating is a tension ripe for study.

Why do we classify signals in this way? Some argue that the ability to represent counterfactuals is critical to all causal reasoning (Sloman & Lagnado Reference Sloman and Lagnado2015; Weisberg & Gopnik Reference Weisberg and Gopnik2013). The attentional and mnemonic advantages of negative emotions that Menninghaus et al. emphasize foster engagement with signals_NDT despite the fact that these signals cannot literally inform. This engagement allows tuning of a vitally important faculty well into adulthood, when pretend play often cedes to propriety.