The ontological state of artworks denotes the mode of being in which artworks are present in our world. Ontologically, it is possible to describe artworks as physical items or as mental items (Figure 1). However, both of these notions do not sufficiently characterize artworks. Therefore, it is necessary to describe artworks as mixed items meaning that they are, in part, physical items and, in part, mental items (Figure 1). Following this notion, Schmücker (2003) denotes artworks as intersubjective-instantial type entities (German translation: “intersubjektiv-instantiale Typ-Entitäten”). This term can be defined as follows: Artworks are entities of a specific class (or type) of items that exist as mental states in a group of people (intersubjective) and, simultaneously, have (or have had) a physical representation in the real world (instantial). In order to refer to an item as artwork, three essential conditions must be fulfilled: (1) There has to be a specific far-ranging consensus on the item that classifies it as artwork, (2) there is or there was at least one incident that led to its creation and (3) there is at least one physical manifestation – or there is at least one mental state that guarantees the possibility of its physical manifestation (Schmücker, 2003). The ambiguity of the ontological state is reflected in the intensions of the term artwork.

Since the term artwork escapes a universal definition, the only possibility to define this term is to take a differentiated (philosophical) look at it. Here, it is helpful to discriminate between the extension and the intension of artworks. Extension means that a term is defined by examples. In other words, the term is defined by the range of items to which the term applies (Carnap, 1947). For instance, the extension of the term car consists of all actual cars (BMW 318i, Hyundai i30, etc.). In the case of the term artwork, the extension approach is to great extent related to the assumption that ‘Art is indefinable’ and, therefore, can only be described by examples. Obviously, the extension of the term artwork is not very useful if one wants to define what the essence of an artwork is. Instead, its essence can be described with the intension of the term. Intension implies that a term is defined by its inner meaning. The intension relates to the properties that an object needs to have in order to be counted as referential to the term. In other words, an intensional definition provides the meaning of a term by specifying necessary and sufficient conditions for when the term should be applied. Therefore, the intension aims at the internal concept that constitutes a formal definition. Referring to the example term car, the intension would be that a car (1) is a self-powered motor vehicle that (2) is used for transport, (3) is a product of the automotive industry, (4) runs primarily on roads, (5) has seating for one to eight people and (6) has four wheels. These six intensions of the term car apply to all or at least most cars. This is the case for most non-philosophical terms.

Artwork, however, is an ambiguous term. For centuries, philosophers and scientists from different fields argued over the different intensions of the term artwork (Kant, 1790; Zimmermann, 1865; Bell, 1914; Adorno, 1970; Tatarkiewicz, 1972; Danto, 1981; Gaut, 2005). The intensions can be paraphrased as certain points of view on a specific item (i.e., artwork) and include essentialism, production-aesthetic intentionalism and reception-aesthetic intentionalism, mental functionalism, historicism, institutionalism and the cluster account. Here, it is not the aim to argue for one specific intension, to investigate the intensions in detail or to pinpoint their advantages and disadvantages. Instead, I will focuses on the problems that arise in scientific research if two or more intensions of the investigated item are applied in a mixed fashion. The central issue is that – in contrast to cars – artwork cannot be described with every intension of the term.

In everyday life, it is not problematic to apply more than one intension of the term artwork to one specific item. When somebody visits an art gallery, the classification of an object as artwork is not of practical relevance. In scientific research, however, the application of more than one intension within one study is highly problematic because an exact interpretation of results is impossible without control over the experiment. This is true for the selection and usage of the stimuli in particular.

Following the conviction that artworks are physical objects, they must possess a materialistic medium within our world. This assumption leads to the essentialistic approach (see Figure 2A). The main statement of the essentialism (or formal objectivism) is that there are formal (perceivable) object properties that are universal to any artwork (Fechner, 1876; Bell, 1914; Osborne, 1952; Redies, 2015). Most essentialistic hypotheses aim at beauty and not at the decision whether an object is an actual artwork. At least until the late 19th century, art had been closely associated with beauty (Tatarkiewicz, 1972). Nowadays, several research groups focus on formal aspects of beauty and found, for instance, preferences for a specific ratio of organization divided by complexity (Birkhoff, 1933), for curved over sharp objects (Bar and Neta, 2006), for symmetry (Jacobsen et al., 2006), and for self-similarity (Redies et al., 2012). However, preferences for certain object properties cannot define whether an item is an artwork or not. The complex ontological state of artworks – they are no simple physical objects, but also have an inherent mental component – opposes this assumption. Nevertheless, universal properties of artworks are widely discussed.

In contrast to essentialism, in which artworks are considered as purely physical objects, intentionalism also takes into account mental aspects, that is, artworks are considered as mixed items. In this view, artworks are man-made items that transmit emotions and/or other non-verbal information (see Figure 2B). Usually, the (supposed) intention of the artist is considered to be detached from an influence on the perceiver. Therefore, intentionalism can be divided into two types: (1) The production-aesthetic intentionalism and (2) the reception-aesthetic intentionalism. In the production-aesthetic intentionalism, the artist‘s intention is communicated through the artwork. Danto (1981) argues that the intention of the artist to produce an artwork and to decide what is expressed in this particular artwork is of utmost importance. According to him, it is impossible to perceive an artwork without an interpretation of it. This credo also applies to the reception-aesthetic intentionalism in which an artwork is an item that has an influence on the perceiver. In visual aesthetics, participants’ reactions to artworks are frequently measured (Augustin et al., 2011; Leder et al., 2013). With the application of reception-aesthetic intentionalism, an investigation of inter-individual differences between participants is possible as well. In this body of research, personality traits have been linked to preference differences for art styles (Furnham and Walker, 2001) and specific types of abstract artworks (Lyssenko et al., 2016).

The mental functionalism also considers artworks as mixed items. To some extent, mental functionalism is related to the reception-aesthetic intentionalism. The main difference is that functionalism is not focusing on individual perceivers, but rather on general functions of artworks. From a functionalistic point of view, an artwork is a social construct that has to fulfill a specific purpose (e.g., as a symbol, see Figure 2C). The term artwork, therefore, applies to items that represent something. For a visual artwork to be a symbol, it does not necessarily have to resemble an original item in order to function as iconic representation (Goodman, 1968). Instead, it can also exemplify particular properties. Pale colors, for instance, can represent sadness or loneliness. Therefore, exemplification is a kind of interpretation – but an interpretation that can be followed quite easily. Partly in line with this, Adorno (1970) proposes a dialectic approach, according to which every critically relevant artwork criticizes society and is the source of utopian power. Therefore, the artwork – and Adorno emphasizes especially the form, not the content – relates to history and societal relationships. According to him, artworks illustrate the state of the society and, in particular, its decadence.

Historicism can be seen as a special version of functionalism. In historicism – in contrast to Adorno’s assumptions – items get the status of artworks only if they relate to previous artworks (see Figure 2D) – and not necessarily to society, as it is the case in functionalism. Carroll (1999) claims that we identify an item as an artwork if it is possible to embed the item in a contingent story in relation to previous items that have already been identified as artworks. Though artworks are described as items that show whatsoever similarity to other artworks (at least for most of the topics), the mental reflection about these items is of equal importance. Similar to intentionalism and functionalism, artworks are seen simultaneously as physical objects and as mental items (mixed items) in historicism.

The actual physical form of appearance (or manifestation) of an artwork is completely negligible in institutionalism (see Figure 2E). For institutionalists, an artwork is an item that has been nominated as an artwork. Dickie (1974) proposed that an item is an artwork (1) if it is a candidate for appreciation and (2) if it is presented to the art world by authorized representatives. In this scenario, the institutionalism itself functions as a kind of black box, in which some item is put and, as a result, there is an evaluation whether it is an artwork or not (Piecha, 2003). This process can be rather arbitrary. It can, however, also be based on some of the other intensions mentioned above because the representatives (or “experts”) might evaluate items according to essentialistic, intentionalistic, historic or functionalistic criteria. While this is certainly the case for some artworks, the actual process of evaluating the artwork often remains rather obscure in most cases.

The cluster account as proposed by Gaut (2005) combines all previously presented intensions (see Figure 2F). Following it, there is no generally favorable concept of the term artwork. Instead, items can gain the status of artworks by exhibiting certain physical properties by their communicating intentions, their influence on the perceiver, their symbolic function, their relation to art history, their simple nomination or by a combination of these intensions. Supporters of the cluster account declare that it is impossible to categorize every approved artwork (following the extension of the term) by one of the mentioned intensions. After an item has been denoted as artwork, the justification for this denotation is investigated. A major problem is the post hoc characteristic of this approach. By identifying an item as an artwork, it automatically receives a new state in the world. A subsequent interpretation, as proposed by supporters of the cluster account, necessarily depends on this newly acquired state (Piecha, 2003).

Referring to their ontological state, artworks are mixed items (Schmücker, 2003). If an object is perceived as an artwork, it possesses a specific inherent mental component and, therefore, the perception mode differs between artworks and non-artworks. This issue has been addressed in a study on disgusting objects. When items were claimed to be artistic, affective responses by the participants were more positive (Wagner et al., 2014). Furthermore, labeling IAPS (International Affective Picture System) images as artworks reduces positive emotional reactions (Gerger et al., 2014). In other words, we look differently at an object when we classify it as an artwork. This might be based on the expectation of deeper meaning, which may lead to a more intensive look at the object (Danto, 1981). For this reason, inferring from experiments on general perception to perception of art, and vice versa, is highly problematic. Of course, it would be unfair to say that one can never draw conclusions on general perception from results that stem from experiments, in which artworks are used as stimuli. But for all that, researchers should be very careful when assigning conclusions from mixed items to materialistic items.

As stated before, artworks (mixed items) are perceived in a way different from non-artworks (materialistic items). Marin and Leder (2016) showed that complexity ratings after long exposure time differed between photographs of environmental scenes and representational art paintings. Possibly, cognitive interference in art perception (controlled processing) that is related to associations and interpretation attempts might influence the judgments (Marin and Leder, 2016). Furthermore, several studies showed that the confrontation with artworks can activate the so-called default mode network (a network of interacting brain regions that is usually active when the focus is not on the outside world). Interestingly, the activation of this network is mediated by different artworks in different participants (Vessel et al., 2012, 2013). The default mode network is not activated when participants perceive non-artistic stimuli (materialistic items). Therefore, if researchers want to draw conclusions on art perception from experiments on non-art stimuli (e.g., Muth and Carbon, 2013), an elaborate discussion is inevitable. Only then, it might be possible to consider a transfer of the results.

A related problem occurs when researchers alter images of artworks. In a study on ambiguity, Jakesch et al. (2013) investigated paintings by René Magritte. They used two different versions of 36 Magritte paintings: an original version and a photoshopped version, in which the inherent ambiguity had been removed. The authors stated that “the original and the manipulated version only differed in terms of ambiguity […]” (Jakesch et al., 2013, p. 5). However, this statement is not entirely correct. The images did not only differ in terms of ambiguity but also in terms of their ontological state. Consequently, the authors compared ambiguous images of artworks (i.e., mixed items) with non-ambiguous images that were no artworks (i.e., materialistic items). A similar problem can be found in several other studies (McManus et al., 1993; Di Dio et al., 2011). It has been shown that digital images that are thought to be created by the researcher are less appealing than digital reproductions of artworks (Kirk et al., 2009). Most importantly, images that have been artificially created by scientist are – in most cases – not works of art. Though they are solely generated to fulfill a specific purpose (i.e., to function as stimuli in an experiment), they do not qualify by any intension to be categorized as artworks. An exception is the reception-aesthetic intentionalism. Possibly, participants perceive these images as works of art. However, if this is assumed, the researcher has to control for participants’ perception mode. One possibility is to ask participants post hoc if they think they saw artworks as stimuli and, therefore, control for perceived authenticity (Pelowski et al., 2017). Perceived authenticity influences the evaluation of a painting’s quality as well as the artist’s talent (Wolz and Carbon, 2014). While asking participants whether the perceived stimuli were images of artworks is surely a helpful approach, even if participants state that they got the impression to be confronted with artworks, this does not solve the problem entirely. If the participant perceives a non-art image and – based on his assumption that the image depicts a real artwork – reacts falsely as if this is an image of an artwork, it is not plausible to expect the exact same reaction as if the participant perceives the image of a real artwork. A naïve viewer may be not affected by certain features of artworks (no automatic processing of fluency) but at the same time categorize the stimulus as artwork because of the expectation to be confronted with art and/or his misleading assumptions how art should look like. Nevertheless, a post hoc report by the participants whether they assume that they perceived images of real artworks could be helpful to control the results. In any case, the results of the respective experiments should be interpreted carefully. In other words, researchers should be aware of the specific aesthetic perception mode for artworks.

Another problem based on the ontological state of artworks is the mixed-item problem. Researchers often treat artworks as physical items and neglect their mental component. However, if artworks were simple physical objects, a (perfect) copy of the artwork would be indistinguishable from the original. This is, however, not the case. Therefore, showing pictures of artworks on a computer screen is – in many cases – highly problematic. While it is possible to measure image properties (a perfect copy possesses presumably the same image properties as the original), it is implausible, for instance, to study certain reception-aesthetic intentionalistic effects with these copies. As the same physical reaction are not elicited when we see a photograph of Anne Hathaway or Hugh Jackman than when we meet them for real, it is also not the same to see a picture of an artwork and to be faced with the original. In a study on titles of abstract artworks, for instance, Leder et al. (2006, p. 181) presented computer images of abstract artworks to participants, as it is commonly done in experiments. Then, they asked whether the participants “understood the artist’s intention,” “whether the artwork affected them emotionally,” and what their “thoughts evoked by the artwork” were. Of course, the participants answered these questions. But, in fact, it was illegitimate to ask such questions because no real artworks were present. The aura of the original and, subsequently, the mental state of the respective item makes artworks special. Therefore, representations of at least some artworks will not evoke the same effect in the observer as the originals do. Similar problems can be found in other studies (Vartanian and Goel, 2004; Villani et al., 2015).

The effect of representations of artworks is rather difficult to examine. One has to differentiate between the already mentioned perceived authenticity (Pelowski et al., 2017) and the actual authenticity. Even if authenticity is correctly perceived – meaning that the participant that is confronted with an image of a real artwork reacts as if he is confronted with the real artwork– a copy does not feature the aura and the temporal and spatial uniqueness of the original (Benjamin, 1969). Therefore, the reaction to the copy might be different (Locher et al., 1999). In summary, researchers should be aware that some investigations are illegitimate when performed on mere copies of actual artworks (for details, see section “Control over Stimuli”). Thus, such experiments should be avoided or they should be conducted with actual artworks.

A rather similar problem arises from the location of the experiment. Researchers should be aware of the difference between a museum setting and a laboratory setting. Usually, visitors are in a certain state of mind when visiting a museum (Brieber et al., 2015) which might be related to the prestige of institutionalism in our time (Pelowski et al., 2017). When participating in an experiment, their state of mind is not necessarily the same. For instance, the contour of the room influences ratings on beauty and pleasantness (Vartanian et al., 2013). Additionally, other aspects of presentation context like framing (Redies and Gross, 2013; Ensor and Hamilton, 2014), boundaries (Cupchik, 2006), lighting (Griswold et al., 2013) and size (Pelowski and Akiba, 2011) have been discussed to be relevant for art appreciation. Quite often studies on artworks target on the state of mind of the perceiver. In these cases, the location of the participant has to be considered in order to reach a valid interpretation of the results (Cela-Conde et al., 2011; Gartus and Leder, 2014).

The application of two different intensions for the same items/stimuli in scientific research can lead to erroneous conclusions. This discrepancy becomes problematic, for example, if the selection and the experimental usage of the stimuli do not hold up to the same standards. Two examples of hypothetical cases will demonstrate this issue. Afterward, three actual studies will be described.

In many cases, researchers select their stimuli based on institutionalism. As described above, the selection process of museums could or could not be associated with another intension. If a researcher selects classic and modern artworks based on institutionalism (i.e., they are displayed in a museum), he ignores the fact that the choice of the museum to display them may have been based on various factors. Classic artworks are subject to a different selection process than modern artworks. On the one side, the selection of classic artworks is, presumably, often based on essentialistic and production-aesthetic intentionalistic intensions. On the other side, modern artworks are, presumably, often chosen because of historic, functionalistic and reception-aesthetic intentionalistic intensions. The key point is that researchers are probably not aware of the selection process in most cases. This problem can be denoted as the incalculable cluster account. Every stimulus selection based on institutionalism is problematic.

In an eye movement study on visual interest, Locher et al. (2007) used images of eight paintings by renowned artists from The Metropolitan Museum of Art, New York. The stimuli “were selected to represent a range of styles along the abstract-representational continuum” (Locher et al., 2007, p. 57). As already pointed out above, the selection process for museum paintings is often rather obscure. One could assume that Jan Vermeer’s “Young Woman with a Water Pitcher” (1660–1662) was chosen for its artistic beauty (essentialism) while Fernand Léger’s “Woman with a Cat” (1921) was chosen as a prime example of cubism (historicism). Alternatively, it is conceivable that the paintings were exhibited in the museums just because they were works by famous artists. The point is that we do not know what led to their exhibition. It cannot be ruled out that such decisions have an influence on the results of the study. Several other studies face similar problems (Mallon et al., 2014; van Paasschen et al., 2014; Villani et al., 2015). In contrast to the first three problems, there is no solution for this problem. Researcher in the field of aesthetics should avoid the incalculable-cluster-account problem and not base their stimulus selection on institutionalism.

Researchers should be aware that the selection of the stimuli has an effect on the results. While this principle is basically true for every kind of research, it becomes even more important in aesthetics research because the stimuli are denoted with an ambiguous philosophical term (artwork). In order to produce internally valid (within study) results, the researcher has to select his stimuli according to the purpose of the study.

First, the respective researcher has to decide whether it is required to use actual artworks as stimuli. In general, the usage of images (copies) of artworks is problematic if reactions of participants are measured (reception-aesthetic intentionalism and cluster account, automatic processing). When focusing on the mere physical aspects of the artwork (essentialism), it is sufficient to use copies. Also, studies investigating functionalistic and historistic aspects can be carried out with copies because it is not necessary to see the original artwork in order to understand its symbolism and/or to file it into the art world. In other cases, the respective usage has to be challenged.

Secondly, researchers should decide whether pre-studies for the selection of stimuli are necessary. Pre-studies are especially useful if artworks are considered as mixed items. They are helpful to gain a better control over the stimuli and to match them to the purpose of the study. Concerning stimuli, pre-studies should be used as a tool to create a dataset that is adequate to the aim of the study. For instance, artworks for the pre-study could be selected based on the institutionalism and the pre-study itself filters out the particular artworks that are symbols with a specific purpose (functionalism). Thus, pre-studies can be used to convert an incalculable cluster account to an intension of choice (Augustin et al., 2011). Of course, it is not always required to apply the same intension of the term artwork to the selection of stimuli and their investigation. Instead, every singular study has to be challenged. If the purpose of the experiment is to investigate the difference between two or more intensions (calculable cluster account), several pre-studies should be carried out in order to create according datasets.

Due to the subjectivity of aesthetic experiences, the control over participants is of utmost importance in aesthetics research. Especially when investigating controlled processing, an intensive briefing of the participants (providing participants with information on the stimuli/experiment, leading the focus of the participant, etc.) is necessary because controlled processing largely depends on the state of mind of the participant. In contrast, automatic processing should be largely unaffected by briefing (Graf and Landwehr, 2015).

Additionally, some studies require the acquisition of information on the participants beyond usual data (e.g., gender, age, and educational degree). For instance, degree of art expertise, familiarity with the stimuli, religious beliefs and mood could be of interest. This information is essential in experiments on controlled processing because of the influence of personal characteristics. Also if the application of functionalism is the purpose of the study, the study will benefit from information on the participants. Some functional aspects are processed automatically (e.g., subconscious reaction on colors as representatives for certain moods). This might also be the case for investigations of essentialistic and intentionalistic characteristics. Therefore, prior to these studies, the acquisition of information on the participants should be considered.

One purpose of the decision tree is to display, which researcher decisions may lead to what kind of problems. The possible problems have been elaborated in section “Artworks As Stimuli in Aesthetics Research”. Most of these problems can be avoided if researchers are aware of the challenges that come with their decisions. The decision tree (Figure 5) gives an overview on which problems are likely to occur in which experiments.

Additionally, I strongly encourage researchers to state the intensions that have been applied. In order to conduct externally valid (between studies) experiments, researchers should not only state which intension of the term artwork they focus on in their investigation (usage of stimuli, see Figure 5), but also which intension of the term artwork they applied when they selected the stimuli. Furthermore, studies would benefit from an explanation on why the respective intensions were applied. Only then, the reader of the study will be fully informed.

Aesthetics research is very complex. Every single study faces different challenges associated with the usage of the term artwork. Therefore, it is impossible to provide the reader with exact instructions for specific experiments. Instead, the illustration of the ontological state, the given definitions of the intensions of the term artwork, the reference to the PIA Model and the decision tree are meant to assist future researchers to find their own way to internally valid (within one study) and externally valid (between studies) research.