Beauty and aesthetic pleasure

A representative definition of beauty goes back to the philosopher George Santayana, who defined beauty as being value positive, intrinsic, and objectified.¹² Value positive means that it provides pleasure; intrinsic refers to the fact that this pleasure is immediate, without intermediate reasoning. This notion goes back to the medieval philosopher and theologian Thomas of Aquinas.¹³ Finally, objectified means that the audience attributes the pleasure they experience to the object, unlike some cold drink on a hot day, where pleasure is due to palatability. Beauty can be felt only with a disinterested view on the object of pleasure.¹⁴ This is the situation where the museum visitor looks at a painting and feels immediate pleasure that is attributed to the painting, not to one’s own response. I use the term “beauty” when I refer to an aesthetic object, like a painting or a sonata, and the term “aesthetic pleasure” when I refer to the feeling of a person, but the two terms denote different aspects of the same interaction between person and object.

(p. 225 ) Processing fluency

Processing fluency, or simply fluency, is defined as the ease with which information flows through the cognitive system (which includes both perceptual and conceptual components). Processing is said to be fluent if the flow of information is fast and easy, and disfluent if the flow of information is slow and difficult. This feeling can be measured by asking people about the ease with which they can process information (“How easy is it to see this object?”; “How easy is this word to read?”), or by assessing the speed of the underlying cognitive processes.¹⁵ There are several subcategories of processing fluency; for example, we call perceptual fluency experienced ease while processing input to the perceptual system, and conceptual fluency experienced ease while processing conceptual information. We deal mainly with perceptual fluency in this chapter. However, we shall see later that the ease with which people can process symbolic information from an artwork may also play a role in determining its aesthetic qualities.

Determinants of processing fluency

There are two main determinants of perceptual fluency. First, the architecture of the human mind facilitates processing of specific information, compared to other information. For example, symmetrical patterns are detected more easily than asymmetrical patterns;¹⁶ high figure–ground contrast is perceived more easily than low figure–ground contrast;¹⁷ and shapes with rounded edges are perceived more easily than shapes with jagged edges.¹⁸

Second, prior exposure to an object or event in the environment, in psychology called stimulus, changes the ease with which this stimulus can be processed. If a stimulus is repeated, it can be identified more quickly. Because the object itself does not change when a person is exposed to it, any change in mental processing is necessarily due to changes in the mind of the perceiver. What changes is the ease with which the object can be processed, which is independent of memory accuracy.¹⁹ Exact repetition is not necessary in order to facilitate subsequent processing. People may see examples of sentences that follow the rules of a certain grammar or pictures that follow the rules of an artistic style, such that they are later able to identify whether new sentences follow the grammar or whether new pictures conform to the artistic style they saw previously.

In a seminal study, Arthur Reber showed that people were able to acquire information derived from an abstract grammar from exposure to stimuli that followed the rules of that grammar.²⁰ Specifically, he presented letter strings derived from a finite state grammar, which prescribed legal sequences of letters. Participants had to encode these letter strings. After this training phase, he presented new letter strings that followed the same grammar, and letter strings that did not. Participants had to classify the stimuli into those that followed the grammar and those that did not. Indeed, participants were able to classify the new letter strings at above chance (p. 226 ) without knowing the grammar. This is supposed to be similar to the process of language learning where infants and toddlers learn the rules of grammar without being able to state those rules. In a recent study, novices in dance have been exposed to sequences of classical dance movements that followed a rule. Despite lack of explicit knowledge, novice observers were able to distinguish dance movements as conforming or not conforming to the rule, suggesting that they have implicitly learned the rule behind the dance movement sequences.²¹

In a similar vein, people automatically extract prototypes from exemplars. When exposed to dot patterns that converged on a prototype, participants processed the prototype faster than non-prototypical patterns even when they never saw the prototype before.²² In everyday life, people grow up seeing birds and then build up a representation of a typical bird, which is a prototype. People then classify a robin, which is similar to the prototype, more easily as a bird than for example a penguin.²³ Note that the resulting processing fluency is an interaction between the person and the situation: the person has a certain mental state that facilitates the processing of specific information. This state may emerge from both stimulus attributes (e.g., it is easier to process symmetrical stimuli) and experience with the situation (e.g., it is easier to process information one has encountered before). The situation provides information that fits the mental capabilities of the person to a greater or lesser degree, depending on both the stimulus and the person’s prior experience. The better the fit between stimulus and processing capabilities, the more fluently the stimulus is processed, both in terms of speed of perception and subjective reports of ease of processing.

Consequences of processing fluency

All the attributes listed in the previous section increase the positive evaluation of a stimulus: people like symmetrical stimuli more than asymmetrical ones,²⁴ simple shapes with high contrast more than simple shapes with low contrast,²⁵ shapes with rounded edges more than shapes with sharp-angled or jagged ones,²⁶ repeated visual stimuli more than novel ones,²⁷ letter strings following a rule more than irregular strings,²⁸ and prototypical stimuli more than non-prototypical visual stimuli.²⁹ In the latter study, perceptual fluency, as measured by response times in a classification task, mediated the effect of prototypicality on attractiveness, but there was a direct link from prototypicality to positive affect that fluency could not account for, suggesting that fluency accounted for part but not all variance of the attractiveness judgments.

In all these studies, participants in experiments were asked to judge their affective experience. Such judgments are susceptible to subtle influences, for example the order in which stimuli are given.³⁰ Therefore, researchers rely on psychophysiological methods to bolster questionnaire data by measuring bodily (physiological) processes that tell us something about psychological states. Indeed, psychophysiological findings support the notion that fluent processing is inherently positive. (p. 227 ) In one study,³¹ participants had to judge their positive and negative reactions to line drawings. Fluency was manipulated by presentation duration: the longer the drawings were shown, the more fluently they could be perceived. When the participants looked at the line drawings, activity of the zygomaticus major was measured. Activation of this muscle indicates positive affect because it is contracted when people smile. The result of the study was clear-cut: the more fluently a drawing could be processed, the more the zygomaticus major was activated, thus providing further support for the link between fluency and positive affect.

Why does fluent information processing result in positive affect? High fluency signals to the person that things are familiar and the ongoing cognitive processes are running smoothly,³² whereas difficulty of ongoing processing signals that things are not going well. If fluency signals smooth interaction with the environment, then one would predict that fluency should influence processing style. As long as information can be processed fluently, everything appears to be fine, and the person should be more inclined to process information heuristically. If it becomes difficult to process information, however, this signals that something is going wrong, and a person has to attend closely to the situation, resulting in analytical processing. Indeed, this is exactly what has been found after presenting tasks where a spontaneous solution contradicted the right solution, which could be achieved by thinking harder. Song and Schwarz, for example, asked the question, “How many animals of each kind did Moses take on the Ark?” The spontaneous solution most people provide is two, but thinking harder makes them notice that it was Noah, not Moses, who took the animals on the ark. When the task was written in a difficult-to-read font people solved it better than when the task was written in an easy-read font. Presumably, the disfluent font slowed down ongoing processing and made people think harder, resulting in better performance.³³

We have discussed fluency effects on affect. However, fluency influences variables other than affect, such as judged familiarity or judged truth. In one study on judged truth, we presented statements in the form “Osorno is in Chile” and instructed participants to indicate whether the judgment is true or not.³⁴ We manipulated fluency by whether a statement was shown in a color that contrasted strongly with the background, rendering statements well readable, or in a color that contrasted more weakly, rendering the statement moderately readable. We found that participants judged easily readable statements to be more likely true than moderately readable statements. This finding extended existing research that demonstrated that repeated statements were more likely to be judged true than new statements. Presumably, repeated exposure to statements facilitated their processing, increasing the experience of fluency while re-encoding the statements, and this ease of processing affected judgments of truth. Note that the findings discussed in this section demonstrate that processing fluency influences both positive affect and judgments of truth. This may help explain why mathematicians sometimes equate beauty with truth in that they find beautiful theories intuitively more plausible than theories that lack this sense of beauty.³⁵

(p. 228 ) In sum, fluent processing feels positive because it signals that objects or events are familiar and the interaction with the environment is going smoothly. In addition, it has been shown that fluency has an impact on cognitive judgments, such as judged truth, suggesting that beauty and truth have some common underlying mechanisms. We are now ready to review the central assumptions of the fluency theory of aesthetic pleasure, to discuss some of the empirical evidence in favor of the theory, and to explore fluency and aesthetic pleasure outside the realm of art.

Fluency and aesthetic pleasure

Processing fluency yields positive affect, but how does this positive affect translate into aesthetic pleasure? We made four assumptions.³⁶ First, as we have already seen, objects differ in the fluency with which they can be processed. Variables that facilitate fluent processing include stimulus features, like symmetry, and figure–ground contrast, as well as experience with a stimulus, such as repeated exposure or acquisition of prototypes. Second, fluency is itself hedonically marked, and high fluency results in positive affect. Third, fluency feeds into judgments of beauty because people draw on their subjective experience in making evaluative judgments, unless the informational value of the experience is called into question. This observation can be explained by the so-called affect-as-information framework. When people have to evaluate a stimulus, they may not assess stimulus content, but ask themselves, “How do I feel?” The resulting assessment of one’s own feeling then feeds into the judgment at hand. Thus, if people can process a stimulus easily, this yields a positive affective feeling, which then is attributed to some aspect of the stimulus: its beauty in case of visual stimuli, or truth in case of a statement.

Fourth, we hypothesize that the impact of fluency is moderated by expectations and attribution. Presumably, fluency has a particularly strong impact on affective experience if its source is unknown and fluent processing comes as a surprise, in line with findings about fluency effects on judgments of truth.³⁷ If this were not the case, people would never prefer Fugue No. 1 in C-major by Bach to Twinkle, Twinkle, Little Star. When hearing a fugue from Bach, listeners do not easily recognize its structure, like they presumably do in a nursery rhyme. Therefore, high fluency comes as a surprise when hearing the fugue by Bach, which is predicted to yield special pleasure. On the other hand, the spontaneous, fluency-based affective experience is discounted as a source of relevant information when the perceiver attributes the experience to an irrelevant source. As we have seen, people like visual stimuli more if they encountered them repeatedly. However, if people notice that fluency stems from a salient pattern of repetition, they attribute fluency to repetition, which removes the positive affect due to fluency.³⁸

Evidence for the fluency theory of aesthetic pleasure

There is a lively tradition to study aesthetic pleasure in empirical aesthetics. We discussed earlier two main determinants of processing fluency: stimulus attributes (p. 229 ) and prior exposure. We look at how the determinants of fluency influence aesthetic pleasure when perceiving artworks.

For effects of stimulus attributes, let us go back to the old dream of researchers to find the aesthetic formula. This research began even before the first official laboratory of experimental psychology was founded. In 1876, the German physicist Gustav Theodor Fechner published experiments performed in order to find laws of aesthetic. He presented students with rectangles of different proportions and asked them how beautiful they thought the presented stimulus was. He observed that the Golden Section is the proportion his participants perceived as most beautiful.³⁹

It later turned out that there is no special preference for the Golden Section, and that Fechner’s finding must have been a methodological artifact.⁴⁰ Although Fechner did not find a law of aesthetics, he ignited a search for such a formula that went well into the 20th century.⁴¹ Mathematician George Birkhoff published a book titled Aesthetic Measure in which he approached aesthetics mathematically.⁴² He thought that aesthetic measure is a ratio between complexity and order. In contrast, psychologist Hans-Jürgen Eysenck postulated, based on his own research, that it is not the ratio, but the product of complexity and order: the more complex and the more orderly (e.g., symmetrical) shapes were, the more they are preferred.⁴³ Eysenck concluded that “the pleasure derived from a percept as such is directly proportional to the decrease of energy capable of doing work in the total nervous system, as compared with the original state of the whole system.”⁴⁴ This statement can be seen as an early formulation of the fluency theory of aesthetic pleasure.

There are today good arguments against the notion of an aesthetic formula. It has been shown that the objects that experts prefer are different from those novices prefer, rendering the search for one formula difficult.⁴⁵ Moreover, repeated exposure increases preference for an object without changing its physical features; an aesthetic formula would not cover changes in preference due to familiarity with the object. Therefore, there does not seem to exist “one formula that fits all” in aesthetic preference. Although all attempts at finding the aesthetic formula failed, the collected data showed that novices liked simple patterns,⁴⁶ and a recent study revealed that horizontal and vertical lines can be viewed more efficiently, an effect that contributed to aesthetic preference in Mondrian paintings.⁴⁷ Together with the other stimulus attributes that determine visual preference, like figure–ground contrast, roundedness of surface, and symmetry, this suggests that processing fluency is especially apt to explain the aesthetic preferences of novices. We shall later see how effects of conceptual fluency can explain the more complex preferences of experts.

After having explored stimulus attributes, we turn to the second determinant of processing fluency, prior exposure. As discussed above, the simplest way to make people familiar with an object is to expose them to it; people prefer repeatedly presented objects to new objects they have never seen before, which is the mere exposure effect.⁴⁸ From all determinants of fluency, effects of repeated exposure are the most-studied phenomenon in empirical aesthetics. In the research discussed (p. 230 ) in the introduction to this chapter, preference for paintings and music depended on frequency of exposure.⁴⁹ Recently, researchers showed acquisition of preferences from implicit learning of musical structure.⁵⁰ Although the nature of this learning has remained unclear, it became clear that participants were able to extract information from the musical excerpts that were composed in agreement with the 12-tone technique.

Fluency and aesthetic pleasure outside the realm of art

In the past century, physicist Hermann Weyl worked on a theory of gravitation called Raum-Zeit-Materie. His theory turned out to be wrong, but Weyl thought it was so beautiful that he did not wish to abandon it and kept it alive for the sake of its beauty. Much later, it turned out that Weyl’s theory was useful—not as a theory of gravitation, but as a formal principle that could be incorporated into another theory, quantum electrodynamics. This anecdote suggests that even in rigorous sciences like mathematics and physics, aesthetic considerations can be important to assess a hypothesis or a solution to a problem.⁵¹ Other examples of aesthetic pleasures in everyday life come to mind: looking at a beautiful landscape; children adoring the family’s Christmas tree; a mechanic hearing the sound of a perfectly tuned Harley-Davidson. As philosopher John Dewey noted over 80 years ago, many of our everyday experiences outside art museums and galleries include an aesthetic component: “So extensive and subtly pervasive are the ideas that set Art upon a remote pedestal, that many a person would be repelled rather than pleased if told that he enjoyed his casual recreations, in part at least, because of their esthetic quality.”⁵² Dewey advocated bringing the aesthetic back from museums and galleries to the activities of everyday life. Indeed, applications of the processing fluency theory of aesthetic pleasure influenced theory and research in several scientific disciplines outside empirical aesthetics, including marketing,⁵³ finance,⁵⁴ archaeology,⁵⁵ and the cognitive psychology of mathematical intuition.⁵⁶ One of the promises of the fluency theory lies in the fact that it helps understand aesthetic pleasure observed both inside and outside the realm of art.

Empirical challenges to the fluency theory

Three challenges are going to be discussed in this section. First, some experimental results apparently cannot be explained by the fluency theory. Second, the (p. 231 ) inverted-U–shaped function of complexity on aesthetic pleasure apparently contradicts the notion that fluency influences aesthetic pleasure. Third, how does fluency influence the affective experience of emotionally negative content?

First, two studies revealed that their manipulation of the stimulus attributes influenced affective judgment, but not fluency; consequently, their effects could not be explained by fluency. The first study manipulated whether shapes were rounded or sharp-angled.⁵⁷ The stimuli were shown briefly and participants had to decide whether they liked or disliked the presented stimulus. Fluency was defined as the time it took to make the like/dislike judgment, and this response time measure did not differ between the two stimulus conditions. However, decision latency is inadequate to measure the ease with which rounded and sharp-angled stimuli can be processed; a proper test would be to assess perceptual fluency with some response time measure and with a subjective ease measure.⁵⁸ Indeed, a similar experimental manipulation revealed that rounded shapes were both easier to perceive and were judged more positively than shapes with jagged edges, supporting the fluency account.⁵⁹ The other study manipulated the quality of photographs and again used decision latency for liking ratings to measure perceptual fluency. As the authors found no difference for response times between high-quality and low-quality photographs, they concluded that fluency did not seem to influence the results.⁶⁰ Again, the proper assessment would have been related to the perception of the photograph, not to the affective decision. Of course, attributes other than fluency may influence affective judgments,⁶¹ but only proper measurement of fluency can establish such a finding.

The second challenge is whether a stimulus could be too fluent to be attractive. For example, people may listen to a song dozens of times, but there comes a time when they get bored with it and no longer choose to hear it, even though this song certainly has become highly fluent. Would a fluency theory not predict that we like the same simple stimuli all the time? In one study that addressed this issue, participants saw Turkish words 3, 9, or 27 times. Study lists were long or short, containing few or many words. Of course, 27 presentations of the same word is much more salient when presented within a short list than within a long list. Then, participants were given a list composed of “old” Turkish words that had been presented before, and “new” Turkish words, and instructed to indicate how much they liked each word. It turned out that participants liked old Turkish words more under all conditions, with one exception: when Turkish words were presented for 27 times in a short list.⁶² The authors interpreted their findings as support of their fluency-attribution framework: fluency due to repetition increases liking as long as the source of fluency is unknown. A decrease could be observed only when the source of fluency became salient. Moreover, each Turkish word was underlined during training in order to indicate how to pronounce the word. The authors manipulated fluency by consistency of pronunciation; they found that after 6 months, fluency due to consistency of pronunciation increased liking even in the salient conditions where words were repeated 9 or 27 times; consistency did not (p. 232 ) have an immediate effect. Presumably, participants did no longer think that the fluency they experienced was due to consistency of pronunciation and used fluency for their liking judgments. In sum, people do not base their affective judgments on fluency when they know where fluency comes from. When they do not know its source, fluency is positive.

This finding helps explain why people begin to like more complex artworks. As they hear the same piece of music dozens of times, or have seen the same simple paintings over and over again, people presumably attribute ease of processing to simplicity of the artwork, which undermines the affective experience. When viewers get familiarized with more complex music or paintings, it is unlikely that they attribute fluency to simplicity, at least for some time, so that fluency due to familiarity breeds positive affect.

The problem of boredom due to familiarity is related to the often-observed inverted-U–shaped curve that denotes aesthetic preference as a function of complexity:⁶³ preference increases from simple stimuli to stimuli with medium complexity and then decreases as the stimuli turn more complex. If fluency per se caused the effect, one would expect that simple stimuli are preferred most, and then preference decreases linearly with increasing complexity. There is an explanation, however, that is in line with the research on attribution of fluency in repeated exposure.⁶⁴ When complexity is low, the source of fluency is salient, and fluency is attributed to the simplicity of the stimulus. As complexity increases, the salience of the source of perceptual fluency decreases, and this enhances the perceived beauty of the object. Further increases in complexity will eventually reduce processing fluency, leading to a decrease in perceived beauty. These mechanisms would combine to form a U-shaped relation between complexity and beauty, as predicted and found by Berlyne and others. We have seen that there is some evidence for such attribution mechanisms, but to date there are no studies that show these effects in artworks.

The third challenge is the affective content of stimuli. It is conceivable that processing fluency increases positive affect in neutral stimuli and maybe positive stimuli, but what about negative stimuli? It may well be that fluent negative stimuli yield more negative feelings and are therefore aesthetically less pleasurable than disfluent negative stimuli. Research findings are scarce and mixed: one study on mere exposure showed that people liked negative stimuli more after repeated exposure,⁶⁵ whereas others found a decrease in preference.⁶⁶ In addition, prototypical negative stimuli, such as prototypical guns, are judged as being more attractive than guns that do not correspond to a prototype.⁶⁷ Further research may reveal the limiting conditions for beauty in negative stimuli. It may of course be that negative stimuli, like guns or spiders, look more beautiful when they can be processed fluently. However, this does not necessarily mean that people like them more. Moreover, extremely disgusting stimuli may be so negative that fluency manipulations cannot further influence the affective reactions unless the stimuli are shown at the limits of visibility, which usually is not the case in research on fluency and affect.

(p. 233 ) We discussed three empirical challenges. First, fluency has to be measured accurately when researchers want to exclude fluency as the mediating mechanism. Second, the fluency theory can explain boredom after numerous exposures and the inverted-U–shaped function of complexity on preference. Finally, it remains to be seen whether fluency is always positive or could lead to more negative evaluations of negative stimuli. Some other challenges remain. For example, not much empirical research has been done on the fluency theory of aesthetic pleasure with artworks as stimuli.⁶⁸ It is desirable that any theory that makes claims about aesthetic experience is supported by empirical evidence that includes artworks as stimuli.

Potential alternatives to fluency

We are going to discuss two questions. First, is fluency nothing more than expectation? Second, could fluency be captured more precisely by inhibition of attention?

Response to music depends on whether a listener’s expectations are fulfilled or violated.⁶⁹ In his book Sweet Anticipation, musicologist David Huron discusses how expectations in music yield affective responses. If listeners can accurately predict the continuation of a piece, they experience positive affect. An unexpected note yields surprise that could have different consequences, some of them negative, like the feeling that the tone is weird, and some of them positive, like laughter or awe, which may be a response derived from fear. According to Huron, surprise itself at first is negative because prediction failed;⁷⁰ however, the subsequent neutral feeling may contrast with the negative consequences of surprise and thus yield mild pleasure. When the predictability of the ensuing note is high, composers can create tension by slowing the tempo. The resolution of the tension then is positive.

Huron’s theory of expectation in music is similar to the fluency theory of aesthetic pleasure introduced above. Could the fluency theory of aesthetic pleasure be discussed in terms of expectations? I do not think so: although Huron includes unconscious expectation, I think there are fluency effects where expectation does not play a role. Let us assume that Alma has once seen a reproduction of The Yellow Cow by Franz Marc in a book about art history. She visits her sister Sophie in her new apartment, and when she enters, she notices that Sophie has a poster of The Yellow Cow on the wall. Had she expected to see the painting, she may have felt pleasure at seeing that her prediction came true. However, Alma may feel pleasure at looking at the familiar painting even if she never had predicted to see the painting in Sophie’s apartment. In cognitive psychology, this is the classical distinction between top-down and bottom-up processing. Top-down means that processing begins with a concept and goes into details. For example, if I know that the object I will see next is a car, this expectation will guide my processing of the details. Had Alma expected to see the painting by Franz Marc, she would have looked for detail (p. 234 ) information in order to confirm this expectation; this is typical for top-down processing. Bottom-up processing begins with the details: I see some elements of an object, like wheels, doors, and windows, and realize that this is a car. In a similar vein, Alma enters the apartment and suddenly sees details of a painting depicting a yellow cow. As she has encountered the painting before, she can process the details faster. In this example, having seen The Yellow Cow by Franz Marc does not create an expectation of what Alma might see in Sophie’s new apartment; Alma, after all, may have seen thousands of other paintings. However, having stored a painting in memory means that whenever Alma sees the painting again, its processing is facilitated, even without having the expectation to encounter it. In my opinion, one would have to extend the definition of expectation in order to accommodate familiarity effects caused by bottom-up processing.

Does a fluency theory explain all effects of expectation? Presumably, expected events always are more fluent. Therefore, there is always a fluency component when explaining effects of fulfilled or disappointed expectation on affect. However, expectations may have cognitive components that are not covered by fluency theory, for example the creation of hypotheses that guide the perceptual process.⁷¹ Huron’s theory goes further than the fluency theory in that he discusses the consequences of surprise and the creation of tension. Whereas the negativity of surprise could be explained by a fluency theory, it remains to be shown whether fluency plays any role in the creation of tension where predicted events are postponed in order to create suspense.

The second theoretical challenge comes from a series of studies by Jane Raymond and her colleagues that supported a so-called attentional inhibition account of positive affect,⁷² which could be seen as a competitor to the fluency theory. In a typical study, participants saw two patterns, a target pattern to which participants had to attend in order to give a response, and a distracter pattern that was ignored.⁷³ Afterwards, participants had to evaluate the distracter patterns and new patterns that had not been previously shown. The mere exposure account discussed above⁷⁴ would predict that the participants would evaluate distracters more positively than new items. In contrast to this prediction, distracter items were evaluated more negatively than new items. Raymond and colleagues argued that inhibition of attention, as in the non-attended distracter items, is affectively negative; that is why it is called the attentional inhibition account. This finding clearly is incompatible with the mere exposure effect—but is it incompatible with a processing fluency account? One could argue that if distracters are processed less fluently than new items, then one would expect distracters to be evaluated more negatively than new items. This is exactly what another study found: distracter items, though repeated, were processed more slowly and were evaluated more negatively than new items.⁷⁵

Although the attentional inhibition account advocated by Raymond and her colleagues is compatible with the fluency theory,⁷⁶ the authors’ research raises an important question: What are the mental and biological mechanisms underlying (p. 235 ) fluency effects on aesthetic pleasure? Fluency is an attractive concept because very different perceptual and cognitive processes contribute to this subjective experience, as discussed earlier. As this experience is grounded in mental and brain processes, it is interesting to know what these are. The attentional inhibition account presented by Raymond and colleagues may be interesting not only as an alternative account that replaces fluency as the variable that affects aesthetic pleasure,⁷⁷ but also as an account at a different level. If it could be shown that processing disfluency is caused by attentional inhibition, it would constitute a mechanism that underlies fluency effects on aesthetic preference. Other perceptual mechanisms relevant to the perception of art may be related to fluency as well.⁷⁸ In a similar vein, neuroanatomical⁷⁹ or neurochemical⁸⁰ accounts of aesthetic preference do not necessarily contradict the fluency theory, but may reflect the same mechanisms at a different level.

Fluency theory and art

Every psychological and neuroscientific theory of aesthetic preference has to meet two challenges. First, is there sufficient empirical evidence that supports the theory? Second, has the theory something to do with art? Whereas scientists are used to meeting the first challenge, they have more problems addressing the second one, as witnessed by criticism from philosophers⁸¹ and art historians.⁸² Although aesthetic experience is not necessarily about art per se, I assume that appreciation of art always includes aesthetic experience of some sort⁸³ and that any aesthetic experience outside art follows the same logic as aesthetic experience related to art.⁸⁴ Therefore, the validity of any theory of aesthetic experience critically depends on whether it is relevant to art. I here outline the challenges for the fluency theory of aesthetic pleasure and summarize recent theoretical work undertaken to meet this second challenge.⁸⁵

A crucial consideration in such a discussion is that the fluency theory of aesthetic pleasure is not about art. George Dickie noted in the 1960s that research in empirical aesthetics, at least as it was conducted at that time, is not relevant to philosophical questions about aesthetics.⁸⁶ As the logic of empirical aesthetics has not changed much, Dickie’s critique is still powerful,⁸⁷ and it parallels the criticism by philosophers⁸⁸ and art historians⁸⁹ of recent neuroscientific work.⁹⁰ Their argument is that empirical findings showing that some works of art have an effect on perception, judgment, or brain activity does not mean that scholars can draw any inferences about art from those observations. The effects one finds are not limited to art, for example. If research shows that processing fluency enhances positive affect, this finding is very general and not specific to art; it applies to any situation where there are differences in processing fluency. For example, lack of processing fluency has been invoked to explain why students do not study mathematical theories that are difficult to understand.⁹¹ Here the relationship between lack of fluency and displeasure is clearly outside the realm of art. Moreover, some (p. 236 ) contemporary theories of aesthetics do not count pleasure as a defining feature of aesthetic experience,⁹² which means that any empirical research on perceived beauty or affective reactions to artworks would not be about art.

A striking feature of modern art is that it is often difficult to understand or process. Nevertheless, some people like it, a phenomenon that directly contradicts the fluency theory of aesthetic pleasure. A revised version of the fluency theory of art is based on the psycho-historical theory of art⁹³ and further developed in Bullot and Reber.⁹⁴ The psycho-historical theory of art basically assumes that artworks are material traces of an artist’s intention and of the historical, social, and cultural context in which the artwork was made. Proper appreciation of art means that the audience relies not on the visible traces of the artwork only, but also on knowledge about the artist’s intentions and the context in which he or she worked. This knowledge, together with the material trace, determines aesthetic understanding.

This aesthetic understanding leads to proficiency with the artist’s intention and the context, which increases fluency due to being better able to integrate the visible traces with what one knows. Fluency does not stem only from ease of perceiving attributes due to perceptual features or prior exposure to the artwork, but from greater semantic coherence due to proficiency, which leads to conceptual fluency. This helps explain the apparent paradox that on average, novices like simplicity whereas experts like more complex works of art.⁹⁵

However, fluency does not have only affective consequences, and paintings often remain disfluent even for a knowledgeable audience. As people like apparently disfluent paintings, the fluency theory of aesthetic pleasure obviously is incomplete. Bullot and Reber discuss two functions of disfluency in paintings: first, the audience may draw inferences from disfluency about meaning with regard to content; second, disfluency may elicit analytical thinking.

Research has shown that people draw inferences from processing fluency, as an experiment by Hyunjin Song and Norbert Schwarz shows. They presented their participants with cooking recipes written in an easy-to-read font or a difficult-to-read font. The authors found that the participants judged the same behavior to take more time, effort, and skill when the print font of the instructions was difficult to read, with adverse effects on the willingness to engage in the behavior.⁹⁶ This finding demonstrates that processing fluency does not only have affective consequences, but also allows drawing inferences.

In a similar vein, the revised fluency theory postulates that perceivers of an artwork draw inferences from fluency and disfluency, and that artists may manipulate fluency as a strategy to express meaning. For example, disfluency in the painting Snow Storm by J. M. W. Turner (Fig. 9.1) allows the inference that it expresses chaos caused by the storm.⁹⁷ In a similar vein, disfluency in paintings by Robert Pepperell expresses uncertainty,⁹⁸ disfluency in paintings by Georg Baselitz meaninglessness.⁹⁹

However, fluency as a cue to meaning is ambiguous: how does a viewer decide whether disfluency in the painting by Turner expresses chaos or meaninglessness? (p. 237 )

As discussed above, the viewer may know some of the artist’s intention, which in Turner’s example is partly given by the title Snow Storm, which presumably is more associated with chaos than with meaninglessness. However, artists may intentionally leave the interpretation of disfluency ambiguous and thus enable multiple interpretations of an artwork.

Moreover, disfluency may make people think. Remember that people think harder when a task is given in a difficult-to-read rather than an easy-to-read font.¹⁰⁰ Decades ago, artists and art theorists like Shklovskij and Brecht must have had the same intuition when they introduced what is known as the alienation effect.¹⁰¹ Brecht argued that the audience in theater automatically identifies itself with the characters in the play. This identification without reflection prevents the audience (p. 238 ) from seeing the reality as it is. To make the audience aware of the reality behind the situation depicted in the play, automatic identification has to be disrupted in order to make people reflect about what is going on. Brecht called this Verfremdungseffekt, often translated as alienation effect, and elaborated on such effects in Chinese drama and Flemish painting. By making a situation strange, the stream of events becomes disfluent, and people begin to analyze the situation. From a fluency point of view, alienation makes an artwork more difficult and thus less pleasurable; the pleasurable quality can be regained only if the artist can put into effect Brecht’s dictum that the difficulties have to be resolved in the end and processing of the artwork as a whole has to be made easy.¹⁰²

In sum, the psycho-historical theory of art allows adapting psychological theories in a way that they become relevant for art. Painters used disfluency strategically to provide expressive meanings that the audience can infer from disfluency. Moreover, knowledge about the historical context and the intentions of an artist increases the fluency with which the artwork can be experienced, and that, in turn, increases aesthetic pleasure even for perceptually disfluent paintings.

It is not only the fluency theory that can be adapted in that way. Research on the role of emotion in art has made some relevant contributions by using methods that take the artistic context of an artwork into account. Paul Silvia reported a study on reading an abstract poem in which the knowledge of the otherwise naïve participants was manipulated.¹⁰³ Participants’ artistic understanding increased, and so did interest in the poem. In another study, Shigeko Takahashi asked art students to produce abstract line drawings that express certain emotion-related concepts, such as anger, joy, tranquility, or human energy. She then selected those drawings that students thought expressed a certain emotion best. Other students then had to rate the drawings and the related emotion words on a semantic differential, which assesses people’s opinion on whether an object possesses certain attributes, such as Beautiful–Ugly, or Active–Passive. Although the students were not given any information about the underlying history of the drawings, and their relation to the emotion words were not mentioned, the students gave ratings that showed surprising conceptual agreement between drawings and the respective words.¹⁰⁴ Here, it has been shown that artists can communicate artistic intention by expressive means in their drawings. The two studies show that it is possible to do psychological research that takes artistic context into account.