Pupillary Responses in Art Appreciation: Effects of Aesthetic Emotions

Lars Kuchinke, Sabrina Trapp,and Arthur M. Jacobs

Freie Universitat Berlin

Helmut LederUniversity of Vienna

The authors examined effects of aesthetic emotions in art appreciation. Subjects were presented threegroups of slides of cubistic paintings that differed in their processing fluency. In an explicit classificationprocedure, subjects were asked to indicate by button press the moment when they recognized anydepicted object in the painting. The time to recognize a depicted object was shortest for high processingfluency paintings, which were also rated higher in their preference. This is in accordance with the“hedonic fluency model” that predicts higher processing fluency being associated with positive aestheticemotions in art appreciation (Reber, Schwartz, & Winkielman, 2004). In addition, higher processingfluency was associated with increased pupil dilations following the point of explicit classification. Thefinding of higher pupil dilation associated with easy-to-process stimuli is interpreted as reflecting aspectsof aesthetic emotions that follow explicit classification of art stimuli as proposed in the “model ofaesthetic appreciation and aesthetic judgments” (Leder, Belke, Oeberst, & Augustin, 2004).

Keywords: aesthetic emotions, pupillary responses, fluency, cubistic paintings

Although empirical aesthetics have a long history in experimen-tal psychology starting with the works of Fechner (1876), theprocesses underlying art appreciation are still not well understood.In his psychophysics approach Fechner (1876) mainly examinedaesthetic experiences triggered by simple geometric stimuli aimingto confirm general laws of stimulus features, such as certainproportions, to evoke aesthetic pleasure. In contrast, recent studiesin empirical aesthetics often focus on the perceiver characteristicsthat determine aesthetic experiences in the persons who are appre-ciating art. As such, results of modern experimental psychologyrather support the Kantian subjectivistic view that art provokesaesthetic experiences in the perceiver (see Kawabata & Zeki,2004). Appreciation of art is seen as an ongoing elaboration ofmeaning (Cupchik, Shereck, & Spiegel, 1994; Leder et al., 2004).For example, Leder at al. (2004) proposed a stage “model ofaesthetic appreciation and aesthetic judgments” that incorporatesongoing cognitive and affective evaluation and mastering at dif-ferent levels of processing, from early perceptual analyses tocognitive mastering and its evaluation. The proposed outcomes ofthis model are aesthetic judgments and the experience of aestheticemotions.

Aesthetic experiences, in particular experiences of preferenceand beauty, are often reported as responses to art (Berlyne, 1971;Leder at al., 2004; Martindale & Moore, 1988; Russell, 2003) andhave been described as a result of evaluation or appraisal processes(Kawabata & Zeki, 2004; Reber et al., 2004; Russell, 2003;Scherer, 2001; but see Silvia & Brown, 2007, for a discussion on

negative aesthetic emotions). For example, additional informationlike a descriptive or elaborative title have been shown to increasethe pleasantness ratings of a painting (Russell, 2003) or its under-standing (Leder, Carbon, & Ripsas, 2006). Such findings supporttheories that claim a close relationship between the pleasure de-rived from looking at a picture and “having grasped its meaningand understood it” (Russell, 2003, p. 100). Thus, part of thepleasure derived from looking at paintings is ascribed to successfulcognitive processing.

Different theories of aesthetic experiences address this issue.According to Berlyne’s psychobiological approach, experiencesare driven by two opposing affect systems, a positive rewardand a negative aversion system (Berlyne, 1971). Aestheticpreference is directly related to physiological arousal followingan inverted u-shaped function. Apprehending art increasesarousal, and it is claimed that the perceiver experiences either adrop or a sudden increase in arousal as pleasurable. Suchhedonic reduction of arousal can be expected as a result ofsolving the challenges provided by the complexity or meaning-fulness of an artwork (Berlyne, 1971). This model has beencriticized by Martindale and Moore (1988), who found noevidence of an inverted relationship between preference andarousal. Rather, they pointed to the role of prototypicality forpositive aesthetic experiences of preference and liking. Proto-typical objects are preferred in everyday life (Rosch, 1975), andprototypicality is an important determinant of preference rat-ings (Hekkert, Snelders, & van Wieringen, 2003; Martindale &Moore, 1988; Winkielman, Halberstadt, Fazendeiro, & Catty,2006).

Generally, some researchers proposed that the cognitive easewith which information is processed leads to positive affect. Thishas been discussed as an effect of processing fluency as part of the“hedonic fluency model” (Reber et al., 2004; Winkielman &Cacioppo, 2001; Winkielman, Schwarz, Fazendeiro, & Reber,2003). This hypothesis assumes that affective feedback during

Lars Kuchinke, Sabrina Trapp, and Arthur M. Jacobs, Department ofPsychology, Freie Universitat Berlin, Berlin, Germany; Helmut Leder,Department of Psychological Basic Research, University of Vienna, Vi-enna, Austria.

Correspondence concerning this article should be addressed to LarsKuchinke, Freie Universitat Berlin, Department of Psychology,Habelschwerdter Allee 45, 14195 Berlin. E-mail: kuchinke@zedat.fu-berlin.de

Psychology of Aesthetics, Creativity, and the Arts © 2009 American Psychological Association2009, Vol. 3, No. 3, 156–163 1931-3896/09/$12.00 DOI: 10.1037/a0014464

156

cognitive evaluation is used to internally monitor cognitive pro-cessing. A positive affective response toward fluent processing isexpected to indicate good progress toward the goal of successfulmastering, which is itself rewarding and supposed to play a mo-tivational role for further processing steps (Winkielman &Cacioppo, 2001). The authors presented evidence that processingfluency, which varied as a function of the visibility of line draw-ings, affected preference ratings as well as corresponding reactionsin facial electromyography. Easy-to-process pictures elicitedhigher preference ratings and higher activations over the region ofthe zygomaticus major, which is known to indicate positive affect(Winkielman & Cacioppo, 2001). Similarly, Reber et al. (2004)reviewed data that suggest that processing fluency of aestheticobjects is closely associated with aesthetic pleasure. For example,subjects were found to report higher liking ratings for concretepictures following conceptual priming, which was proposed toincrease the processing fluency of the related concepts.

While in most of the previous studies the stimuli used wererather relatively simple, it remains an empirical question whethera sudden understanding of a rather complex, and even ambiguousstimulus such as an artwork might reveal something about thegenuine nature of aesthetic experiences. Because artworks oftenrequire some perception and cognitive processing until a state ofambiguity is resolved (Leder et al., 2004), we used cubist artworksof different classes of accessibility (Hekkert & Van Wieringen,1990) to measure the relationship between the processing of highand low fluency paintings and the aesthetic response. Becausethere is yet not much evidence for the affective consequence offluency through understanding, we measured pupillary responseson artworks.

Pupillometric Indices During Art Appreciation

So far, empirical studies in aesthetics have mainly relied onbehavioral data, like response times and preference ratings.Recently, psychophysiological measures have been introducedin the field of empirical aesthetics to further evaluate aestheticexperiences (e.g., Hofel & Jacobsen, 2007; Jacobson & Hofel,2003; Lengger, Fischmeister, Leder, & Bauer, 2007) and theaesthetic emotions associated with processing fluency (Wink-ielman & Cacioppo, 2001).

We believe that the use of psychophysiological data, inparticular measuring pupillary responses, could provide newevidence for the affective processes involved in art apprecia-tion. Task-evoked pupillary responses have reliably been shownto be as sensitive to affective (Beatty, 1982; Mudd, Conway, &Schindler, 1990; Steinhauer, Boller, Zubin, & Pearlman, 1983;Vo et al., 2008) as to cognitive aspects of processing (Beatty &Kahneman, 1966; Granholm, Asarnow, Sarkin, & Dykes, 1996;Just & Carpenter, 1993; Kuchinke, Vo, Hofmann, & Jacobs,2007). The peak pupil dilation has been found to vary with theamount of cognitive load associated with a memory task (Beatty& Kahneman, 1966) or emotional processing (Hess, 1965;Janisse, 1974). Regarding its role in emotional processing, twoopposing proposals have been made. While Hess (1965) dis-cussed the hypothesis of bidirectional pupillary responses toemotional stimuli with pupil dilations when looking at positivepictures and pupil constrictions for negative material (also seeMudd et al., 1990), others found pupil dilations in response to

emotional stimuli independent of their actual valence (Janisse,1974; Steinhauer et al., 1983; Partala & Surakka, 2003; Vo etal., 2008). Because the later studies controlled for the initialpupil constriction following stimulus presentation while thefirst two did not, these diverging results might partly be attrib-uted to methodological differences. Taken together, these re-sults support the assumption that pupil dilations are associatedwith the resources allocated to the processing of emotionalstimuli.

Pupillary responses to artworks were measured by Libby,Lacey, and Lacey (1973). They observed that unpleasant slidesproduced greater pupil dilation than pleasant slides, which bothdilated less than neutral slides. Moreover, the most reliablefinding on pupil dilations in Libby et al.’s (1973) study was aneffect of the artworks’ interestingness. The more interesting orattention-getting an artwork was rated, the greater the pupildilated—independent of its’ complexity and arousal.

The “model of aesthetic appreciation and aesthetic judg-ments” by Leder et al. (2004) proposes that aesthetic emotionscontribute differently to art appreciation at certain processingstages. Early perceptual analyses of art stimuli could be expe-rienced as less emotional than later stages of explicit classifi-cation and evaluation. Thus, the perceptual stages provide thematerial that reveals emotions in the later processing stages inrespect to cognitive mastering, interpretation, and understand-ing. Therefore, to reveal an effect of aesthetic emotions, weimplemented a task in which we directly focused on the momentof recognition of depicted objects during the apprehension of anartwork—that is the point of explicit classification (Leder et al.,2004). Subjects were asked to immediately indicate the momentwhen they recognized one or more familiar objects depicted inthe cubistic painting (Hekkert & van Wieringen, 1990). Usingthis approach, Hekkert and van Wieringen (1990) observed thatthe time to recognize the depicted objects successfully con-structed sets of artworks, which systematically differ in the easewith which the depicted objects can be recognized. We assumethat this ease is equivalent to the processing fluency of theartworks. Hence, the ease of processing or processing fluencyof these artworks should be reflected in the time needed tosuccessfully recognize their depicted elements. Furthermore, inaccordance with expectations from preferences for representa-tive art (Leder et al., 2004), we also expected easy-to-processpaintings to elicit higher preference ratings.

Using the explicit classification approach, we measured boththe time needed to recognize a depicted object and the pupillaryresponses following the point of explicit classification of thecubistic paintings. By varying the processing fluency of thepaintings as a function of the content accessibility or abstract-ness of the paintings, we expected that the processing advantageassociated with processing fluency is reflected in the explicitclassification times (Reber et al., 2004). In addition, pupillaryresponses were expected to differentially monitor the aestheticemotions associated with processing fluency. In contrast toLibby et al. (1973) where subjects passively viewed artworks,the pupillary responses around the point of explicit classifica-tion were expected to be modulated by aesthetic emotions. Inparticular, higher processing fluency was expected to elicithigher pupil dilations following the point of explicit classifica-tion of depicted objects reflecting enhanced aesthetic emotions.

157PUPILLARY RESPONSES IN ART APPRECIATION

Method

Subjects

Twenty-five female and 10 male students and art novices (M !27.5; SD ! 3.8) from Freie Universitat Berlin participated in thisstudy. The participants gave informed consent and received 5! fortheir participation. Data from eight participants were lost becauseof technical problems during pupil recording. The remaining 27participants had normal or corrected-to-normal vision and reportedno history of neurological and affective disorders.

Stimulus Material

To test our hypotheses, a one-factorial design was appliedcomprising the within-subject factor processing fluency (high,medium, low). Thirty-nine colored slides of cubistic paintingswere selected from a larger list of 60 cubistic paintings previouslyrated on a 5-point scale on complexity, ranging from “low” (1) to“high” (5); abstractness, “representational” (1) to “abstract” (5);and familiarity, “unfamiliar” (1) to “highly familiar” (5), by agroup of 66 students from the University of Vienna. The 60cubistic paintings were divided into three lists with increasing butnonoverlapping rated abstractness. Thirteen slides of each list wereselected so that these three experimental lists differed significantlyin respect to their mean abstractness (in the following calledcontent accessibility), but did not differ in terms of rated familiar-ity (an analysis of variance (ANOVA) revealed no significantdifferences for the rated familiarity across the three content acces-sibility conditions, p ! .288, see Table 1).

The slides with the highest content accessibility (the list with thelowest abstractness ratings) were assigned to the high processingfluency level, medium content accessibility to medium processingfluency level, and the lowest content accessibility slides to the lowprocessing fluency level (see Figure 1 for examples of the cubisticpaintings). The selected paintings depicted either human figures orlandscapes. Special effort was taken to control for the luminance ofthe slides of paintings, known to affect the pupillary response(Loewenfeld, 1999). The picture slides were edited using AdobePhotoshop and the colorlab toolbox for MATLAB to be compa-rable on mean picture size and mean luminance (both ps " 0.180;Table 1). To ensure that participants tried to recognize objects theycould not anticipate, five cubistic paintings were selected as fillerslides that depicted other objects than human figures and land-scapes. The filler slides were edited to be comparable to theexperimental stimuli on mean size and mean luminance.

Procedure

The experiment took place in a medium-illuminated laboratory(background luminance about 500 lux). Subjects seated themselvesin a comfortable chair with their head stabilized in a chin rest(approximately 0.6 m distance between their eyes and the com-puter screen). Pupil data were recorded with a video-based IViewX Hi-Speed eye tracker (SensoMotoric Instruments, Teltow, Ger-many). An infrared sensitive camera recorded pupil diameters at asampling rate of 240 Hz. The experimental session started withwritten instructions on the computer screen, followed by the cal-ibration of the subjects’ left eye. Subsequently, six training trialswere presented that were not part of the stimulus material to assure

that the subjects understand the instructions correctly. The stimulisubtended at maximum a vertical visual angle of 17.19° and ahorizontal visual angle of 15.28°.

The 44 slides (39 experimental and 5 filler slides) were ran-domly assigned on the computer screen using Presentation 9.0software (Neurobehavioral Systems, Albany, Canada) includingthe IVIEW X interface to synchronize trial presentation and pupildata recording. A single trial started with the presentation of afixation cross (#) in the center of the screen for 1,000 ms. Thefixation cross was replaced by an experimental slide that remainedon the screen until button press with a maximum trial duration of9,000 ms. The slide remained on the screen for 1,000 ms followingthe button press for a continuous pupil recording around the timeof the response (see Figure 1d).

Subjects were instructed to press the mouse button as quickly asthey had recognized a concrete object depicted in the painting. Inrandom intervals a control question appeared on the screen fol-lowing a trial, where subjects were asked what they had seen in thepictures. These questions were included in the procedure to assurethat the subjects correctly followed the instructions and respondedonly when they satisfyingly identified a recognizable object.1 It isimportant to note, that no mention was made to the subjects thatonly figures and objects of landscapes were potential recognizableobjects in the experimental slides—a fact that was further sup-ported by the use of our filler slides that included other contents.

Following each trial, the presentation of a smiley indicated a3,000-ms period where subjects were allowed to blink before thenew trial started with the presentation of a fixation cross. The testphase lasted approximately 8 min during which pupil raw datawere recorded. Following the test phase, subjects were given aquestionnaire with all 39 experimental slides printed in high qual-ity to rate them on a 5-point scale on complexity, “low” (1) to“high” (5); abstractness, “representational” (1) to “abstract” (5);familiarity, “unfamiliar” (1) to “highly familiar” (5); and prefer-ence, “do not like” (1) to “highly prefer” (5).

Data Preparation and Analysis

Pupil data were prepared using a computer algorithm written inMATLAB (Version 6.5) that discarded trials with major blinks orlinearly interpolated smaller artifacts on a trial by trial basis in thetime window between 200 ms before stimulus onset and 10,000 mspoststimulus onset. Raw pupillary raw data were sampled down toa 60 Hz and smoothed using a 7-point weighted average filter. Inaddition, all trials were checked visually for undetected artifacts.Because of excessive blinking or recording artifacts during thelong interval, 15.4% (ranging from 0% to 53.8% per subject) of alltrials were discarded. Pupillary artifacts were not systematicallydistributed across experimental conditions.

A response-based analysis was computed where the baselinepupil diameter was defined as the average pupil diameter recordedduring the last 200 ms preceding the subjects’ response and sub-tracted from the raw pupil diameter. Response-based peak dila-tions (PDs) were computed as the maximum baseline-corrected

1 Examples of participants’ answers for depicted objects in the presentedpaintings are: “a face” or “a person” in Figure 1a, “houses” or “a bridge”(Figure 1b), and “houses” or ’boats’ (Figure 1c).

158 KUCHINKE, TRAPP, JACOBS, AND LEDER

pupil diameter in the 1,000-ms interval following the subjects’responses. PDs were computed for all trials and afterward aver-aged per condition and subject. The response-based peak pupildiameter was expected to reflect affective processing around thepoint of explicit classification.

Only the pupillary responses to experimental trials were con-sidered for the statistical analyses. Average PDs per experimentalcondition and subject were submitted to a one-way repeated mea-sures ANOVA comprising the within-subject factor “processingfluency” (high, medium, low). Significance level was set at $ !.05, and a Greenhouse-Geisser correction was applied if necessary.

Results

To check whether the processing fluency manipulationshowed the expected effect in the behavioral data, two repeatedmeasures ANOVAs on response times and preference ratingswere computed. The analysis of the response times revealed amain effect of processing fluency, F(2, 52) ! 126.671, p %.001; &2 ! 0.830. The mean response times showed a paramet-ric pattern with shortest response times when processing cub-istic paintings with high processing fluency and the longestresponse times to paintings with the lowest processing fluency(see Table 2, Figure 2). A series of Bonferroni-corrected t tests

revealed that all pair wise comparisons were significant (allps % .001). The analysis of the preference ratings revealed amain effect of processing fluency, F(2, 52) ! 8.858, p % .001;&2 ! 0.254. The mean response times showed a parametricpattern with highest preference ratings to high processing flu-ency paintings and with lowest preference ratings to paintingswith low processing fluency. A series of Bonferroni-corrected ttests revealed that the preference ratings differed between thehigh processing fluency condition and both the medium and thelow processing fluency condition (both ps % .009).

The response-based PD showed a main effect of processingfluency, F(2, 52) ! 4.795, p ! .012; &2 ! 0.156 (see Figure 2).The Bonferroni-corrected t tests revealed a parametric effect in thepupil dilation following the explicit classification with the highestPD in the high processing fluency condition, medium in themedium processing fluency condition, and lowest in the low pro-cessing fluency condition (see Figure 3). The follow-up pairwise ttests reached significance threshold only for the high versus lowcontrast ( p ! .006, Bonferroni corrected), whereas all other com-parisons were not significant.

Additional analyses of correlation further verified these results.Item-based correlation coefficients averaged across the subjectswere computed between the mean pupillary responses and the

Figure 1. Sample slides of high (1a; “Femme et pot de moutarde, ”Picasso, 1910), medium (1b; “Le reservoir(Horta d'Ebre),” Picasso, 1909), and low (1c; “Fishing Boats,” George Braque, 1909) processing fluencypaintings presented in the experiment and (d) the experimental procedure (see text).

Table 1Stimulus Characteristics as a Function of the Processing Fluency Conditions

Processing fluencyNumber of

slides FamiliarityContent

accessibility Complexity Size (in mm2)Luminance(in cd/mm2)

High 13 2.71 3.00 3.02 200.12 156.46Medium 13 2.71 3.25 3.84 200.15 159.55Low 13 2.57 3.69 4.30 196.20 162.40

159PUPILLARY RESPONSES IN ART APPRECIATION

mean postexperimental ratings (complexity, abstractness, familiar-ity, preference) for each cubistic painting. The results showed thatthe pupil dilation following the subjects’ responses (PD) wassignificantly correlated with mean abstractness ratings (r !(0.551; p % .001) and mean preference ratings (r ! .333; p !.038), whereas no significant correlations were observed withcomplexity (r ! (0.234) and familiarity (r ! .022). Thus, higherpupil dilations following the subjects’ responses were associatedwith higher preference ratings (a measure of aesthetic pleasure)and with lower abstractness ratings (which we assume reflecthigher processing fluency).

Discussion

In this study, participants were asked to indicate the point ofexplicit classification of recognized objects depicted in abstract cub-istic paintings. We found evidence for a relation between the pupillaryresponse and aesthetic pleasure. Following the point of explicit clas-sification, the highest pupil dilations were found for high fluencystimuli and the lowest pupil dilations for low fluency stimuli. More-over, a post hoc analysis of correlation revealed that the pupil dilationfollowing the subjects’ behavioral response was positively related toindividually rated preference of the cubistic paintings.

In the present study, cubistic paintings with high contentaccessibility showed the shortest response times, indicating aneffect of processing fluency. Moreover, the behavioral datasupported the assumption of the hedonic fluency model thathigher processing fluency is also associated with higher pref-erence ratings (Winkielman & Cacioppo, 2001). Although ourdata present evidence for a relationship between content acces-sibility and processing fluency, one might ask whether content

accessibility is a good operationalization of processing fluency.Stimulus features like prototypicality (Hekkert & Wieringen,1990) and stimulus complexity (Bornstein, Kale, & Cornell,1990) have been shown to contribute to preference judgments.However, from a theoretical point of view, it is difficult toisolate the factors that contribute to processing fluency (seeWillems, van der Linden, & Bastin, 2007, for a review). More-over, recent results strengthen the role of the subjects’ inter-pretation of the nature of fluency, especially in situations wherefluency is unexpected (Whittlesea & Price, 2001; Willems etal., 2007). Because we controlled the familiarity of the cubisticpaintings across the processing fluency conditions, we relateour findings to the content accessibility manipulation that di-rectly influenced the ease of processing of our stimuli. Still,more research is needed on this issue.

A recent study by Locher, Krupinski, Mello-Thoms, and Nodine(2007) found that a gist of a painting, that is, initial informationconcerning its style, compositional aspects, and its semantic mean-ing, can already be reported after a 100-ms glance. Because thisstudy differed in many methodological aspects from the presentone, the results are not fully comparable. The eight paintings inLocher et al.’s (2007) study were presented tachistoscopically,thereby filling a square of 178 cm (in contrast to a computer screenin the present study). Most important, the paintings were mainlyrepresentational. On the one hand, it seems obvious that thesepresentation conditions may have affected the fluency with whichthese paintings could be processed, and on the other hand, Locheret al. (2007) cannot rule out the possibility that subjects’ responsesrelied on the processing of mental images—and thus are not purelyautomatic.

high medium low

Resp

onse

Tim

es (i

n m

s)

0

500

1000

1500

2000

2500

3000

high medium low

PD (i

n m

m)

0,10

0,15

0,20

0,25

Figure 2. Response times (in ms, left) and peak pupil dilation (PD in mm, right) in the 1,000-ms intervalfollowing explicit classification.

Table 2Behavioral and Pupillary Responses as a Function of the Processing Fluency Conditions

Processing fluency

Response times (ms) Peak dilation (mm) Preference ratings

M SD M SD M SD

High 1338.27 480.48 0.211 0.122 3.14 1.08Medium 1919.25 699.75 0.191 0.092 2.87 1.11Low 2650.02 805.29 0.176 0.102 2.80 1.21

160 KUCHINKE, TRAPP, JACOBS, AND LEDER

Still, we agree with their assumption that initial pictorial infor-mation is processed automatically at early (perceptual) processingstages, e.g., stages 1 to 3 in the Leder et al. (2004) model.Moreover, Locher et al. (2007) presented evidence that pleasing-ness ratings following a 100-ms glance (Experiment 1) and anunrestricted viewing time (Experiment 2) are correlated, but that avalid judgment of aesthetic pleasure (resulting in higher pleasing-ness ratings) is only available in the unrestricted viewing condi-tion. This result fits well to the assumption that aesthetic emotionsevolve during the early processing stages and are only fully avail-able at later stages of evaluation and cognitive mastering (Leder etal., 2004). Thus, supporting the idea that aesthetic emotions arebest measured when explicit pictorial information is available(e.g., following the point of explicit classification).

The psychophysiological results add further evidence to theliterature that the pupil dilates not only in response to the amountof cognitive processing, but that pupil dilation also indicates af-fective processing (Beatty, 1982). While we observed the shortestresponse times to high processing fluency paintings and the long-est response times to low processing fluency paintings, the oppo-site pattern was found in the pupil data. If pupil dilation wasrelated to cognitive processing only, one would have expected thatthe highest pupil dilation is observed in the low processing fluencycondition; that is, the condition with the longest response times.This phenomenon is often reported in pupillometric research(Nuthmann & van der Meer, 2005; Raisig, Welke, Hagendorf, &van der Meer, 2007). However, this is not the case in the presentstudy. In fact, the highest pupil dilation following the point ofexplicit classification was observed in the high processing fluencycondition. Accordingly, we interpret this as a result of affectiveprocessing following the explicit classification in the appreciationof artworks (Leder et al., 2004).

However, it is unclear how much negative emotions in responseto the apprehension of the cubistic paintings contribute to the

present results. Silvia and Brown (2007) discussed different neg-ative aesthetic emotions in an expanded appraisal model of aes-thetic experience. Following this model, the dilation of the pupilobserved in the present study could partially be attributed to theprocessing of negative emotions. However, this question cannot befully answered by means of pupillometric research. It has beenshown that the pupil dilation reflects the resources associated withvalence-independent processing of emotional stimuli (Janisse,1974; Libby et al., 1973; Partala & Surakka, 2003). However,processing fluency is typically associated with positive experience(Reber et al., 2004; Winkielman & Cacioppo, 2001). Our analysisof correlation supports this view, in that preference ratings werepositively correlated with PD. We would therefore propose thathigher pupil dilation following explicit classification of objectswith high processing fluency stimuli is not only attributed toaesthetic emotion, but to positive aesthetic emotions in particular.

Still, it is discussed that fluency might be interpreted as inher-ently positive as long as no additional information is available,whereas in other situations (e.g., under prior exposure conditions)the interpretation of fluency is shifted according to its learnedvalidity (Unkelbach, 2006). All stimuli were presented twice in thepresent study, during the explicit recognition phase and in thefollow-up questionnaire. The pupillary responses have only beenmeasured during the first presentation, so that repeated presenta-tions cannot be the basis of the observed differences in the pupil-lary responses in the present study. Nonetheless, a cognitive in-terpretation of the pupil data contradicts the expected processingdemands (with higher demands in the low processing fluencycondition) so that we think that the pupillary response followingthe explicit classification reflects positive aesthetic experiences.

The “model of aesthetic appreciation and aesthetic judgments”(Leder et al., 2004) discusses two different outcomes of art appre-ciation processes: aesthetic judgments and the experience of aes-thetic emotions. Both outcomes are discussed to follow a multi-

Time (in ms)

-1000 -500 0 500 1000

PD (i

n m

m)

-0,2

-0,1

0,0

0,1

0,2

highmediumlow

Figure 3. Response-locked mean pupillary responses as a function of the processing fluency conditions. Thedashed vertical line represents the response.

161PUPILLARY RESPONSES IN ART APPRECIATION

staged evaluation approach. In the present study, the pupillaryresponse indicated aesthetic experiences after the explicit classifi-cation of depicted objects in cubistic paintings. According to themodel, aesthetic emotion depends on the subjective success of theinformation processing (Leder et al., 2004). The present pupillo-metric measures further support the notion of (early) affectiveresponses in art appreciation even if the processes of cognitiveevaluation and appraisal are not fully completed. Additional stud-ies are required that address the question of the temporal dynamicsof affective and cognitive responses using psychophysiologicalvariables with a higher temporal resolution.

In sum, the results of the present study indicate that affectiveaesthetic responses in art appreciation somehow depend on theease with which an aesthetic stimulus can be processed. Cubisticpaintings with high content accessibility are processed faster to-ward the point of explicit classification of the depicted content.This processing fluency advantage is accompanied by two addi-tional findings. First, the pupillary response following this explicitclassification reflects the processing advantage with higher pupildilations to stimuli with higher processing fluency, which can bestbe explained as indicating positive aesthetic emotions. Second,higher processing fluency was found to elicit higher preference ofthese paintings. The results are in accordance with the predictionsof a model of aesthetic appreciation and aesthetic judgments, andit will be interesting to further investigate whether the presentfindings generalize to other aesthetic stimuli like for instancerepresentational artworks or the music domain.

References

Beatty, J. (1982). Task-evoked pupillary responses, processing load, andthe structure of processing resources. Psychological Bulletin, 91, 276–292.

Beatty, J., & Kahneman, D. (1966). Pupillary changes in two memorytasks. Psychonomic Science, 5, 371–372.

Berlyne, D. E. (1971). Aesthetics and psychobiology. New York: Appleton-Century-Crofts.

Bornstein, R. F., Kale, A. R., & Cornell, K. R. (1990). Boredom as alimiting condition on the mere exposure effect. Journal of Personalityand Social Psychology, 58, 791–800.

Cupchik, G. C., Shereck, L., & Spiegel, S. (1994). The effects of textualinformation on artistic communication. Visual Arts Research, 20, 62–78.

Fechner, G. T. (1876). Vorschule der Asthetik. Leipzig, Germany: Breit-kopf & Hartel.

Granholm, E., Asarnow, R. F., Sarkin, A. J., & Dykes, K. L. (1996).Pupillary responses index cognitive resource limitations. Psychophysi-ology, 33, 457–461.

Hekkert, P., Snelders, D., & van Wieringen, P. C. W. (2003). “Mostadvanced, yet acceptable”: Typicality and novelty as joint predictors ofaesthetic preference in industrial design. British Journal of Psychology,94, 111–124.

Hekkert, P., & van Wieringen, P. C. W. (1990). Complexity and proto-typicality as determinants of the appraisal of cubist paintings. BritishJournal of Psychology, 81, 483–495.

Hess, E. H. (1965). Attitude and pupil size. Scientific American, 212,46–54.

Hofel, L., & Jacobsen, T. (2007). Electrophysiological indices of process-ing aesthetics: Spontaneous or intentional processes? International Jour-nal of Psychophysiology, 65, 20–31.

Jacobsen, T., & Hofel, L. (2003). Descriptive and evaluative judgmentprocesses: Behavioral and electrophysiological indices of processing

symmetry and aesthetics. Cognitive, Affective & Behavioral Neuro-science, 3, 289–299.

Janisse, M. P. (1974). Pupil size, affect and exposure frequency. SocialBehavior and personality, 2, 125–146.

Just, M. A., & Carpenter, P. A. (1993). The intensity dimension of thought:Pupillometric indices of sentence processing. Canadian Journal of Ex-perimental Psychology, 47, 310–339.

Kawabata, H., & Zeki, S. (2004). Neural correlates of beauty. Journal ofNeurophysiology, 91, 1699–1705.

Kuchinke, L., Vo, M. L-H., Hofmann, M., & Jacobs, A. M. (2007).Pupillary responses during lexical decisions vary with word frequencybut not emotional valence. International Journal of Psychophysiology,65, 132–140.

Leder, H., Belke, B., Oeberst, A., & Augustin, D. (2004). A model ofaesthetic appreciation and aesthetic judgments. British Journal of Psy-chology, 95, 489–508.

Leder, H., Carbon, C. C., & Ripsas, A. L. (2006). Entitling art: Influenceof title information on understanding and appreciation of paintings. ActaPsychologica, 121, 176–198.

Lengger, P., Fischmeister, F. P. S., Leder, H., & Bauer, H. (2007). Func-tional neuroanatomy of the perception of modern art: A DC–EEG studyon the influence of stylistic information on aesthetic experience. BrainResearch, 1158, 93–102.

Libby, W. L., Jr., Lacey, B. C., & Lacey, J. I. (1973). Pupillary and cardiacactivity during visual attention. Psychophysiology, 10, 270–294.

Locher, P., Krupinski, E. A., Mello-Thoms, C., & Nodine, C. F. (2007).Visual interest in pictorial art during an aesthetic experience. SpatialVision, 21, 55–77.

Loewenfeld, I. E. (1999). The pupil: Anatomy, physiology and clinicalapplications. Boston: Butterworth and Heinemann.

Martindale, C., & Moore, K. (1988). Priming, prototypicality, and prefer-ence. Journal of Experimental Psychology: Human Perception andPerformance, 14, 661–670.

Mudd, S., Conway, C. G., & Schindler, D. E. (1990). The eye as musiccritic: Pupil response and verbal preferences. Studia Psychologica, 32,23–30.

Nuthmann, A., & van der Meer, E. (2005). Time’s arrow and pupillaryresponse. Psychophysiology, 42, 306–317.

Partala, T., & Surakka, V. (2003). Pupil size variation as an indication ofaffective processing. International Journal of Human-Computer Studies,59, 185–198.

Raisig, S., Welke, T., Hagendorf, H., & van der Meer, E. (2007). Investi-gating dimensional organization in scripts using the pupillary response.Psychophysiology, 44, 1012–1016.

Reber, R., Schwartz, N., & Winkielman, P. (2004). Processing fluency andaesthetic pleasure: Is beauty in the perceiver’s processing experience?Personality and Social Psychology Review, 8, 364–382.

Rosch, E. (1975). Cognitive representations of semantic categories. Jour-nal of Experimental Psychology: General, 104, 192–233.

Russell, P. A. (2003). Effort after meaning and the hedonic value ofpaintings. British Journal of Psychology, 94, 99–110.

Scherer, K. R. (2001). The nature and study of appraisal: A review of theissues. In K. R. Scherer, A. Schorr, & T. Johnstone (Eds.), Appraisalprocesses in emotion: Theory, methods, research (pp. 369–391). NewYork: Oxford University Press.

Silvia, P. J., & Brown, E. M. (2007). Anger, disgust, and the negativeaesthetic emotions: Expanding an appraisal model of aesthetic experi-ence. Psychology of Aesthetics, Creativity, and the Arts, 1, 100–106.

Steinhauer, S. R., Boller, F., Zubin, J., & Pearlman, S. (1983). Pupillarydilation to emotional visual stimuli revisited. Psychophysiology, 20, 472.

Unkelbach, C. (2006). The learned interpretation of cognitive fluency.Psychological Science, 17, 339–345.

Vo, M., L-H., Jacobs, A. M., Kuchinke, L., Hofmann, M., Conrad, M.,Schacht, A., et al. (2008). The coupling of emotion and cognition in the

162 KUCHINKE, TRAPP, JACOBS, AND LEDER

eye: Introducing the pupil old/new effect. Psychophysiology, 45, 130–140.

Whittlesea, B., & Price, J. (2001). Implicit/explicit memory versus analytic/nonanalytic processing: Rethinking the mere exposure effect. Memory &Cognition, 29, 234–246.

Willems, S., van der Linden, M., & Bastin, C. (2007). The contribution ofprocessing fluency to preference: A comparison with familiarity-basedrecognition. European Journal of Cognitive Psychology, 19, 119–140.

Winkielman, P., & Cacioppo, J. T. (2001). Mind at ease puts a smile on theface: Psychophysiological evidence that processing facilitation elicitspositive affect. Journal of Personality and Social Psychology, 81, 989–1000.

Winkielman, P., Halberstadt, J., Fazendeiro, T., & Catty, S. (2006). Pro-totypes are attractive because they are easy on the mind. PsychologicalScience, 17, 799–806.

Winkielman, P., Schwarz, N., Fazendeiro, T., & Reber, R. (2003). Thehedonic marking of processing fluency: Implications for evaluativejudgment. In J. Musch & K. C. Klauer (Eds.), The psychology ofevaluation: Affective processes in cognition and emotion (pp. 189–217).Mahwah, NJ: Erlbaum.

Received April 4, 2008Revision received July 7, 2008

Accepted July 8, 2008 !

163PUPILLARY RESPONSES IN ART APPRECIATION