The role of similarity, sound and awareness in the appreciation of visual artwork via motor simulation

The role of similarity, sound and awareness in the appreciation of visual artwork via motor simulation

Cognition 137 (2015) 174–181 Contents lists available at ScienceDirect Cognition journal homepage: www.elsevier.com/locate/COGNIT Brief article Th...

772KB Sizes 1 Downloads 23 Views

Cognition 137 (2015) 174–181

Contents lists available at ScienceDirect

Cognition journal homepage: www.elsevier.com/locate/COGNIT

Brief article

The role of similarity, sound and awareness in the appreciation of visual artwork via motor simulation Christine McLean a, Stephen C. Want b, Benjamin J. Dyson c,⇑ a

Ontario College of Art and Design, Toronto, Canada Ryerson University, Toronto, Canada c University of Sussex, UK b

a r t i c l e

i n f o

Article history: Received 26 May 2014 Revised 8 January 2015 Accepted 11 January 2015

Keywords: Embodied cognition Motor action Artwork Congruency Expertise Awareness

a b s t r a c t One way to increase art appreciation is to create congruency between the actions performed by the artist and the actions performed by the viewer. Leder, Bar, and Topolinski (2012) successfully created such a link by asking participants to make either stroking or stippling motions while viewing stroke-style and pointillist-style paintings. We carried out a direct replication of Leder et al. (2012) in Experiment 1 but failed to reproduce their results. In Experiment 2, we achieved the desired cross-over interaction between image and action but only when the relationship was made more transparent. Experiment 3 demonstrated that this effect requires a motor component and cannot be reproduced by simply hearing the sounds associated with drawing production. Experiment 4 investigated whether either an external manipulation or a self-report measure of awareness of the image–action match modulated the liking ratings, in addition to artwork familiarity and participants’ own hypotheses regarding the direction of the image–action effect. Participants who predicted that congruent relationships between what they saw and what they did would increase liking showed enhanced congruency effects. The links between historical production and contemporary exposure to art may then be an overt rather than covert process. Ó 2015 Elsevier B.V. All rights reserved.

1. Introduction The idea that cognition is influenced by the physical constraints of the organism (embodied cognition; Freedberg & Gallese, 2007) is an appealing and popular thesis. Two recent papers have made remarkable claims regarding the role of bodily movements in cognition in the context of aesthetic appreciation. Specifically, when viewing artwork, if viewers produce hand movements that approximate the methods by which the artwork was ⇑ Corresponding author at: Department of Psychology, School of Psychology, Pevensey Building, University of Sussex, Falmer BN1 9QH, UK. Tel.: +44 (0)1273 876661; fax: +44 (0)1273 678058. E-mail address: [email protected] (B.J. Dyson). http://dx.doi.org/10.1016/j.cognition.2015.01.002 0010-0277/Ó 2015 Elsevier B.V. All rights reserved.

originally created (Taylor, Witt, & Grimaldi, 2012), such movements increase liking for the art (Leder, Bar, & Topolinski, 2012). These examples, in addition to resonating with the embodied cognition idea, also connect with how the repetition of action serves to facilitate processing (Hommel, 2004; Pashler & Baylis, 1991). This mimicry of the mode of artistic production at the time of viewing has been claimed to serve almost as a time travel device, creating empathy and ‘‘sympathetic resonance’’ (Leder et al., 2012, p. 1479) between viewer and artist, yielding a deeper appreciation of the work in question. To wit: ‘‘. . .the mere viewing of paintings engages the observers’ motor system. Motor simulations give artists the ability to reach out to their audience across great distances and even generations via paint and canvas. . .’’ (Taylor et al.,

C. McLean et al. / Cognition 137 (2015) 174–181

2012, p. 36), and, ‘‘. . .artistic style as a concretization of the artist’s act of creation links the artwork’s creation with the moment of art perception; this connection bridged 100 years in the present case. . .’’ (Leder et al., 2012, p. 1480–1481). In the latter study, participants in the critical experimental group made either stroking or stippling motions while viewing artwork of a stroke-style and of a pointillist-style. Increases in artwork liking were observed when the artistic style resembled the motion produced (congruent; e.g., stroking motion with stroke-style painting) relative to when the artistic style did not resemble the motion produced (incongruent; e.g., stippling motion with stroke-style painting). There are a number of reasons to be cautious about these specific claims. First, it is not currently clear how exact the relationship between art production and art perception has to be. In the case of Taylor et al. (2012), brushstrokes indicative of left-to-right and right-to-left motion appeared to facilitate similar directional movements made by the participant. In the case of Leder et al. (2012), liking of the artwork of Vincent van Gogh and Claude Monet could be enhanced by the request to make 20 cm horizontal stroking motions, while the appreciation of the artwork of Paul Baum and Georges Seurat was facilitated with the request to make stippling or tapping motions. Not only does this appear to do somewhat of a disservice to the original artists’ techniques but it also raises the issue of how aware participants must be of the link between artwork and motion before such an effect is manifest. Second, it is not currently clear how robust these effects are. In the case of Taylor et al. (2012), it appears that the speeding of reaction time when the direction of the brushstroke in the artwork and the direction of actual responding by the participant were congruent was more robust when making a right-to-left as opposed to a left-to-right response. In the case of Leder et al. (2012), the average magnitude of the difference between artwork liking when congruent and incongruent actions were undertaken while viewing the artwork was 0.55 on a 1–7 scale. This appeared equivalent to an effect observed in the opposite direction, favoring incongruent action when viewing pointillist paintings, for a control group ( 0.46) where action and viewing were temporally distinct (although see Ticini, Rachman, Pelletier, & Dubal, 2014, for the apparent success of a temporally distinct action-viewing effect using pointillist-style paintings). Therefore, to investigate the specificity and robustness of motor simulation effects on art appreciation, we began with a direct replication of Leder et al. (2012) with higher statistical power. 2. Experiment 1 2.1. Methods 136 undergraduate students from Ryerson University, Toronto, Canada, consented for their data to be used in analysis and no observations were excluded. As per a recent recommendation regarding the sample size of replication attempts (Simonsohn, 2013), we chose a sample size

175

of at least 2½ times the size of the original experiment. We focused on replicating Leder et al.’s (2012) critical experimental condition, which included 48 participants, split between a stippling group (n = 24) and a stroking group (n = 24). Our stippling group (n = 70) consisted of 57 females with mean age of 19.5 years (sd = 4.6). The stroking group (n = 66) consisted of 57 females with mean age of 19.8 years (sd = 5.7). Participants were tested in groups of up to 6. All studies were approved for testing by the Research Ethics Board of Ryerson University, and participants received course credit for participation. The study began with participants reading on-screen instructions and filling out their age, gender and years of formal art training on a sheet of paper provided. The 5 examples of ‘‘neoimpressionist, pointillist-style paintings and . . .postimpressionist, stroke style paintings’’ (Leder et al., 2012, p. 1479; hereafter ‘pointillist-style’ and ‘stroke-style’) were used in Experiment 1 and these 10 pieces of art were presented in random order. Participants viewed each artwork for 30 s each via a projector screen. When viewing the artworks, participants generated drawings (either stippling or stroking, following the instructions in Leder et al., 2012) using a dominant hand-held pencil, and a blank piece of paper supported by a mouse mat. Individual covered plastic boxes were used to house and shield the drawing from both the drawer and other participants. In between viewing artworks, participants used their non-dominant hand and a second pencil to respond to an artwork pleasantness prompt (1 = do not like at all, 7 = like very much), presented for 15 s. Following artwork rating, participants completed a brief questionnaire based on Leder et al. (2012) regarding their understanding of the task and then were debriefed. 2.2. Results and discussion A mixed-model ANOVA with the between-participants factor of action (stippling, stroking) and the within-participants factor of image (pointillist, stroke) produced a main effect of image only: F(1, 134) = 30.80, MSE = 0.500, p < .001, gp2 = .187. Both the main effect of action (F(1, 134) = 1.33, MSE = 1.169, p = .251, gp2 = .010) and the interaction between action  image (F(1, 134) = 0.02, MSE = 0.500, p = .880, gp2 < .001) failed to reach statistical significance. As seen in Fig. 1a, participants expressed more liking for stroke-style images (4.53) relative to pointillist-style images (4.05), independently of which hand movement they produced. One possible reason for failing to replicate Leder et al. (2012) might be that the connection between the hand movements and the painting style must be explicitly recognized in order to increase liking of the relevant images. Perhaps in our study, participants simply failed to notice the relationships between their hand movements and the style of the paintings they viewed. Therefore, to strengthen the connection between the hand motions participants made and the pictures they viewed, we presented a selection of the drawings generated by participants in Experiment 1 as the artwork in Experiment 2.

176

C. McLean et al. / Cognition 137 (2015) 174–181

Fig. 1. (a) Graph showing the failure in Experiment 1 to replicate the original interaction between image and action in Leder et al. (2012) and further analysis investigating whether the putative interaction in was influenced by the degree of liking (low raters versus high raters). (b) Examples of participantgenerated stroking and stippling drawing used in Experiments 2 and 3. (c) Graph showing the predicted interaction between image and action under transparent conditions in Experiment 2 and further analysis according to degree of liking. (d) Graph showing the insufficiency of passive exposure to the auditory portion of mimicked action to produce the interaction in Experiment 3 and further analysis according to degree of liking. Error bars represent standard error.

C. McLean et al. / Cognition 137 (2015) 174–181

3. Experiment 2 3.1. Method 133 undergraduate students participated in the study. The stippling group (n = 68) consisted of 55 females with mean age of 19.46 years (sd = 2.74). The stroking group (n = 65) consisted of 55 females with mean age of 18.86 years (sd = 1.99). The design of Experiment 2 was identical to Experiment 1, except that the pointillist and stroke-style artworks were now replaced by 5 stippled and 5 stroked drawings that had been produced by participants in Experiment 1 (see Fig. 1b for examples; see Fig. S1 for the complete set).

3.2. Results and discussion A mixed-model ANOVA with the between-participants factor of action (stippling, stroking) and the within-participants factor of image (stippled, stroked) was again run in Experiment 2. This produced a significant interaction between action  image: F(1, 131) = 4.46, MSE = 0.788, p = .037, gp2 = .033, in the absence of main effects of action (F(1, 131) = 0.12, MSE = 2.147, p = .904, gp2 < .001) or image (F(1, 131) = 0.66, MSE = 0.788, p = .418, gp2 = .005). As seen in Fig. 1c, participants who generated stippling motions preferred stippled images and participants who generated stroking motions preferred stroked images. That said, the magnitude of these effects was small (0.31 in favor of stippled images in the stippling group and 0.14 in favor of the stroked images in the stroking group) and follow-up Tukey’s HSD tests of the action x image interaction revealed that no individual cell differed statistically significantly from any other. In contrast to Experiment 1, mimicry of the motions used to produce artwork at the time of viewing in Experiment 2 increased liking ratings to a small degree. There was a reduction in liking for the visual stimuli in Experiment 2 (2.84) relative to Experiment 1 (4.29; t[267] = 13.06, p < .001). One reviewer wondered if this reduction in liking was responsible for the emergence of the action–image interaction in Experiment 2 and that, correspondingly, the higher liking ratings in Experiment 1 prevented the emergence of the same interaction. Although we note there was no evidence of a ceiling effect in Experiment 1 (the means of the liking ratings were all around the middle of the range), if this was the case then we would expect the interaction in Experiment 2 to be stronger in those reporting low rather than high liking ratings. Data were median split for each action group according to the rating of disliking/liking expressed (low-liking stippling [n = 33] < 3; high-liking stippling [n = 33] > 3; low-liking stroking [n = 32] < 2.7; high-liking stroking [n = 30] > 2.7). When rating (low, high) was entered as a second between-participants factor into the analysis, the predicted main effect of liking was observed (F(1, 124) = 300.86, MSE = 0.661, p < .001, gp2 = .708) in addition to a three-way interaction (F(1, 124) = 4.33, MSE = 0.789, p = .039, gp2 = .034). As seen in Fig. 1c, the interaction was driven by high rather than low raters,

177

thereby refuting the disliking hypothesis; given that the interaction in Experiment 2 appeared among those participants who expressed higher, not lower, liking for the images, the lack of a significant interaction in Experiment 1 is unlikely to have resulted from the generally higher liking ratings given to the images in Experiment 1 (as compared to the images from Experiment 2).1 Experiment 2 appears to support the idea that congruent links between action and vision enhance the liking of the visual input (e.g., stippling action–stippled painting), relative to incongruent links (e.g., stroking action–stippled painting). What is clear when producing these actions however is that there is also a significant auditory component: the stippling action produces a distinct ‘tapping’ noise while the stroking action produces a distinct ‘sweeping’ noise. The embodied cognition hypothesis requires that the cross-modal effect (see Spence, 2011, for a review) is between action and vision, not between vision and audition. To rule out the latter account, Experiment 3 asked whether passive exposure to the auditory portion of mimicked action was sufficient in promoting enhanced liking ratings for vision.

4. Experiment 3 4.1. Method 98 undergraduate students participated in the study. The stippling group (n = 49) consisted of 39 females with mean age of 21.24 years (sd = 5.33). The stroking group (n = 49) consisted of 40 females with mean age of 20.41 years (sd = 3.88). Data collection stopped as a result of semester end. The design of Experiment 3 was identical to Experiment 2, except that the request for participants to produce stippling or stroking motion during viewing was replaced by the sound of stippling or stroking motion. Prior to Experiment 3, 10 examples of stippled drawing and sound, and, 10 examples of stroked drawing and sound were recorded in a south-attenuated booth using a Tascam DS-1 Portable Sound Recorder. Sounds were edited down to 30 s clips using Audacity, with 500 ms onset offset amplitude ramps to avoid click, and played back at a comfortable volume over loudspeakers. Sounds are available upon request. 5 stippled and 5 stroked drawings were selected for Experiment 3 (see Fig. S1 for the complete set) and all 10 stippling/stroking sounds were played concurrently with drawing presentation, according to condition, and in random order. 1 At the request of the same reviewer, the data from Experiments 1 and 3 were also median split according to liking for each action (Experiment 1: low-liking stippling [n = 35] < 4.25; high-liking stippling [n = 35] > 4.25; low-liking stroking [n = 32] < 4.4; high-liking stroking [n = 29] > 4.4) or sound group (Experiment 3: low-liking stippling [n = 24] < 2.1; high-liking stippling [n = 23] > 2.1; low-liking stroking [n = 23] < 2.8; high-liking stroking [n = 23] > 2.8). In neither case was the critical three-way interaction significant (Experiment 1: image  action  liking, F(1, 127) = 2.20, MSE = 0.512, p = .141, gp2 = .017; Experiment 3: image  sound  liking, F(1, 89) = 0.42, MSE = 0.631, p = .518, gp2 = .005), suggesting that the (lack of significant) interaction between image or sound and action was not influenced by whether participants expressed relatively low or high liking of the images.

178

C. McLean et al. / Cognition 137 (2015) 174–181

4.2. Results and discussion A mixed-model ANOVA with the between-participants factor of sound (stippling, stroking) and the within-participants factor of image (stippled, stroked) revealed a main effect of sound: F(1, 96) = 5.72, MSE = 1.794, p = .019, gp2 = .056, in the absence of a main effect of image (F(1, 96) = 2.35, MSE = 0.652, p = .128, gp2 = .024) or sound  image interaction (F(1, 96) = 0.82, MSE = 0.652, p = .367, gp2 = .008). As seen in the left-hand panel of Fig. 1d, stroking sounds (2.77) led to increased liking ratings relative to stippling sounds (2.31). The data from Experiment 3 suggest that passive exposure to the auditory portion of mimicked action was insufficient to modulate liking ratings in the predicted fashion. Therefore, the cross-modal effect we observed in Experiment 2 is more likely be an embodied effect; that is, an action–vision effect rather than a visual-auditory effect.

5. Experiment 4 To address the lingering question of why our replication of Leder et al. (2012) did not reproduce the original effect, we explored a number of possibilities in Experiment 4. Leder et al.’s (2012) undergraduate sample from the University of Vienna may have simply contained more art experts than our sample: across our Experiments 1–3 (n = 366), only 42 individuals reported formal art training (average 2.90 years training). An exploratory analysis suggested that the expected interaction was slightly more evident in those individuals reporting formal art training, although conclusive statements were not possible due to the low numbers of such individuals. In Experiment 4, we broadened our definition of formal art training to include pre-university exposure. If expertise contributed to the effect there should be a positive correlation between the number of years of formal art training reported and the magnitude of the action–image congruency effect. The identification of ‘art experts’ though really serves as a proxy for a number of perceptual, memorial, cognitive and motoric differences relative to non-experts. Consequently, we explored what some of these differences might be, specifically, that familiarity differences between the two art styles and awareness of action–image links might contribute to the effect. One explanation for the absence of the action–image interaction in Experiment 1 (and the presence of the same interaction in Experiment 2) concerns potential differences in familiarity with the artworks used (we thank an anonymous reviewer for this suggestion). It is possible that participants in Experiment 1 were more familiar with stroke-style images and consequently rated them as more liked than pointillist-style images. This main effect of image may have washed-out the more subtle interaction between image and action. In contrast, Experiment 2 used entirely unfamiliar images. This may have resulted in both sets of images being (dis)liked to the same degree, such that the absence of a main effect of image allowed the more subtle interaction to express itself. This idea appears contrary to recent data to suggest that art familiarity does

not correlate with liking ratings for pointillist-style paintings in the context of preceding congruent and incongruent motor action (n = 18; Ticini et al., 2014). Nevertheless, the contribution of familiarity to the relationship between action–image congruency and liking was tested by asking participants at the end of the study to also rate their familiarity with both pointillist-style and stroke-style paintings. If familiarity differences between the two types of art work play a role in suppressing the action–image congruency effect, then there should be a negative relationship between the absolute value difference between familiarity with pointillist-style and stroke-style paintings, and the size of the congruency effect between action and visual input. In other words, the action–image congruency effect should emerge most clearly for participants who are equally familiar with the two types of art work. A second hypothesis regarding why action–image congruency effects sometimes emerge (Leder et al., 2012 and our Experiment 2) and sometimes do not (our Experiment 1) was that some form of awareness of the similarity between artwork technique and hand movement leads to the effect. At first blush, the request to make horizontal stroking motions roughly 20 cm in length (Leder et al., 2012, p. 1480), and the delicate characterization of the cliffs of Pourville by Claude Monet do not appear to relate to one another perceptually. One way to appreciate the somewhat oblique relationship between the two is to have explicit knowledge about the production of such works and recognize the family resemblance between the pragmatics of certain painting techniques and the motions performed at the time of viewing (c.f., art-historical context; Bullot & Reber, 2013, p. 126). Again, compared to our Experiment 1, perhaps Leder et al.’s sample contained participants who were more likely to explicitly recognise this family resemblance. To test this idea in Experiment 4, approximately half the participants completed an enhanced awareness version of Experiment 1, in which the relationship (but not the direction) between motor movement and artwork style was pointed out prior to liking judgments. The approximate other half of participants completed a standard awareness condition, which was identical to Experiment 1. Participants were also asked after the experiment whether they were aware of the link between image and action, to serve as a second, self-report measure of awareness. If awareness (defined either by external instruction or self-report) plays a role in aesthetic decision making, then the size of the congruency effect between action and image should be greater in the enhanced awareness condition relative to the standard awareness condition, and in those individuals who reported being aware of a relationship relative to those individuals who reported being unaware of a relationship. Irrespective of participants’ awareness of the relationship we also asked at the end of the experiment whether they thought any hypothesized relationship between action and image would increase, decrease or have no effect on their liking ratings. If demand characteristics contributed to any observed image–action congruency effects, then we should see enhanced congruency effects in those individuals who believed that congruency between image and action would increase liking of the work.

C. McLean et al. / Cognition 137 (2015) 174–181

179

5.1. Method 242 undergraduate students contributed data in Experiment 4, now consisting of 4 groups. The standard awareness stippling group (n = 64) consisted of 53 females with mean age of 20.47 years (sd = 0.38). The standard awareness stroking group (n = 61) consisted of 51 females with mean age of 20.31 years (sd = 0.37). The enhanced awareness stippling group (n = 57) consisted of 47 females with mean age of 19.39 years (sd = 0.38). The enhanced awareness stroking group (n = 60) consisted of 55 females (1 participant responded n/a to gender) with mean age of 19.40 years (sd = 0.35). Data from 3 additional participants were rejected: 1 for attempting to copy the images, 1 for producing both clear strokes and dots, and 1 for incomplete questionnaire responding. The design of Experiment 4 was identical to Experiment 1, with four critical changes. First, there was a between-participants manipulation of awareness in that approximately half the participants received the same instructions as Experiment 1 (standard awareness), whereas approximately half the participants received additional instructions (enhanced awareness), in which the relationship between the artwork and motion was stressed. Specifically, the extra instruction was: ‘‘The style of each painting – whether it was made by using stroking motions or dotting motions – will either match or nor match the dotting (stroking) action you are performing’’. Second, a more liberal definition of formal art training was supplied to include any pre-university training participants may have had. Third, after participants completed their liking ratings, they were asked to rate their familiarity with the two styles of artwork on a 1–7 scale (1 = never seen this style before, 7 = seen hundreds of example of this). Fourth, at the end of the study participants were asked whether – at the time of looking at the artwork – they were aware of a match between half of the artworks and the motions they were asked to produce, and, whether they predicted such a match would increase, decrease or make no difference to their liking ratings. 5.2. Results and discussion With respect to performance in the standard condition (a direct replication of Experiment 1), a mixed-model ANOVA with the between-participants factor of action (stippling, stroking) and the within-participants factor of image (pointillist, stroke) produced a main effect of image only: F(1, 123) = 23.95, MSE = 0.433, p < .001, gp2 = .163. Both the main effect of action (F(1, 123) < 0.01, MSE = 1.074, p = .977, gp2 < .001) and the interaction between action  image (F(1, 123) = 0.02, MSE = 0.433, p = .899, gp2 < .001) failed to reach statistical significance. As shown in the left-hand panel of Fig. 2a, participants showed a general preference for stroking-style painting (4.57) relative to pointillist-style painting (4.16) as per Experiment 1. A similar two-way ANOVA conducted for the enhanced awareness condition also revealed a main effect of image (F(1, 115) = 13.19, MSE = 0.539, p < .001, gp2 = .103) in the absence of a main effect of action (F(1, 115) = 0.64, MSE = 1.619, p = .426, gp2 = .005) and in

Fig. 2. (a) Graph showing the comparison of the standard and enhanced awareness conditions in Experiment 4. (b) Scatterplot showing the lack of correlation between familiarity difference between the two artwork styles and the magnitude of congruency effect in both standard and enhanced awareness conditions. (c) Graph showing the comparison of participants who self-reported awareness of action–vision links versus those who did not. (d) Scatterplot showing the lack of correlation between years of formal art training and the magnitude of congruency effect. (e) Graph showing that participants who predicted that the relationship between action and vision would increase their liking of artwork produced a larger congruency effect than those participants who predicted there would be no difference. (f) Graph showing a close-to-significant interaction between action and vision in those participants who expected congruency to increase liking, relative to those who expected congruency to make no difference to liking. Error bars represent standard error.

180

C. McLean et al. / Cognition 137 (2015) 174–181

the absence of an interaction between action  image (F(1, 115) = 1.06, MSE = 0.539, p = .163, gp2 = .017; see right-hand panel of Fig. 2a). Numerically, there was a rise in liking for pointillist-style images among participants who made stipple actions in the enhanced awareness condition but this effect was very small. We suspect this slight modulation is due to the closer action–vision link between stippling motions and pointillist-style images than the relationship between 20 cm stroking motions and strokestyle images. The familiarity data showed overall greater familiarity for stroke-style painting relative to pointillist-style painting for both the standard (4.92 versus 4.34; t[124] = 3.61, p < .001) and enhanced awareness (4.55 versus 4.00; t[116] = 4.19, p < .001) conditions. However, correlations comparing the absolute value difference between familiarity with pointillist-style and stroke-style paintings with the size of the action–image congruency effect ([pointillist-style–stroking-style] for the stipple action condition, or, [stroking-style–pointillist-style] for the stroking action condition) revealed the absence of a significant correlation for both the standard (r = .137, p = .142) and enhanced awareness (r = .009, p = .922) conditions (see Fig. 2b for the resultant scatterplots). Therefore, familiarity does not appear to modulate any image–action congruency effect (see also Ticini et al., 2014). A further test of awareness, this time relying on the selfreport item on the exit questionnaire, also only produced a main effect of image in both the unaware (F(1, 125) = 15.67, MSE = 0.506, p < .001, gp2 = .111) and aware (F(1, 113) = 20.18, MSE = 0.463, p < .001, gp2 = .152) participants (see Fig. 2c), in the absence of any other effects: unaware main effect of action (F(1, 125) = 1.04, MSE = 1.372, p = .309, gp2 = .008), unaware image  action interaction (F(1, 125) = 0.24, MSE = 0.506, p = .627, gp2 = .002), aware main effect of action (F(1, 113) = 1.68, MSE = 1.241, p = .197, gp2 = .015), aware image  action interaction (F(1, 113) = 1.50, MSE = 0.463, p = .223, gp2 = .013). We suspect these data may be related to the specific nature of the question we asked to assess awareness: ‘‘at the time that you were looking at the artworks, were you aware that there was a match between the hand movement you were asked to produce and half of the artworks?’’ (emphasis added). Here, it might have been the case that participants did not experience a connection between all 5 congruent artworks, and hence marked their response ‘no’. The self-report measure of formal art training failed to correlate with the magnitude of congruency effect expressed (r = .036, p = .581, after rejecting one outlier reporting 30 years of formal art training; see Fig. 2d). Finally, we tested whether participants’ own hypothesis regarding the direction of the image–action effect influenced their responding. Fig. 2e shows the average congruency effect for individuals who believed the effect would increase (n = 88; 0.29), decrease (n = 13; 0.08), or make no difference (n = 140; 0.06) to the liking of images where image and action matched. Given the small n and large variance associated with individuals expecting decreased liking, we ran an independent t-test comparing the increase and no difference groups and the difference

was both significant and in the correct direction (t[226] = 2.45, p = .015). This effect is again shown in Fig. 2f in the context of the standard 2  2 design, revealing there to be a close-to-significant interaction term for the expect increase group (F(1, 86) = 3.79, MSE = 0.457, p = .055, gp2 = .042) and no significant interaction term for the expect no difference group (F(1, 138) = 0.24, MSE = 0.500, p = .626, gp2 = .002). Irrespective of participants’ manipulated or selfreported awareness of the relationship between image and action, individuals who reported expecting any match between artworks and their hand motion (especially in the dotting/pointillist condition) to increase their liking of the works showed a larger image–action congruency effect than those individuals who reported that any match would make no difference to their judgments. Experiment 4 then shows a clear role for expectancy in the production of image–action congruency effects when viewing art works.

6. General discussion The current experimental series reveals the conditions under which links between the historical production and contemporary perception of art impact appreciation of the art works. Specifically, our data suggest that for participants to be more likely to report increased appreciation of a type of artwork, they must be actively engaged in congruent motoric reproduction rather than passively exposed to auditory elements of congruent motoric reproduction (Experiment 3). The effect can arise if there is a clear perceptual link between what is being reproduced and what is being viewed (Experiment 2). If a bottom-up link between image and action cannot be established, then the effect can also arise from a top-down influence and participants’ belief that congruent relationships between image and action should increase their liking of what they are viewing (Experiment 4). Regardless of whether the effect stems from clear bottom-up perceptual links or top-down expectancy effects, the influence of hand movements on liking still appears quite small. Therefore, in contrast to the suggestion that covert simulation modulates the evaluation of artwork (Leder et al., 2012), the importance of obvious connections between motor action and visual perception and the individual expectation of a positive relationship between visuo-motor links and artwork liking suggest that the modulation of artwork judgement as a function of motor action can also be very much an overt process.

Acknowledgements B.J.D. is supported by an Early Researcher Award granted by Ontario Ministry of Research and Innovation. Data was presented at the 24th Annual Canadian Society for Brain, Behaviour and Cognitive Science Meeting, 3-5th July 2014, Toronto, Canada. Correspondence should be addressed to: Ben Dyson, School of Psychology, Pevensey Building, University of Sussex, Falmer, BN1 9QH, UK. Email: [email protected].

C. McLean et al. / Cognition 137 (2015) 174–181

Appendix A. Supplementary material Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/ j.cognition.2015.01.002.

References Bullot, N. J., & Reber, R. (2013). The artful mind meets art history: Towards a psycho-historical framework for the science of art appreciation. Behavioral and Brain Sciences, 36, 123–180. Freedberg, D., & Gallese, V. (2007). Motion, emotion and empathy in esthetic experience. Trends in Cognitive Science, 11, 197–203. Hommel, B. (2004). Event files: Feature binding in and across perception and action. Trends in Cognitive Sciences, 8, 494–500.

181

Leder, H., Bar, S., & Topolinski, S. (2012). Covert painting simulation influence aesthetic appreciation of artworks. Psychological Science, 23, 1479–1481. Pashler, H., & Baylis, G. (1991). Procedural learning: 2. Intertrial repetition effects in speeded-choice tasks. Journal of Experimental Psychology: Learning, Memory and Cognition, 17, 33–48. Simonsohn, U. (2013). Small telescopes: Detectability and the evaluation of replication results. Available at SSRN: http://ssrn.com/abstract= 2259879 or http://dx.doi.org/10.2139/ssrn.2259879. Spence, C. (2011). Crossmodal correspondences: A tutorial review. Attention, Perception, & Psychophysics, 73, 971–995. Taylor, J. E. T., Witt, J. K., & Grimaldi, P. J. (2012). Uncovering the connection between artist and audience: Viewing painted brushstrokes evokes corresponding action representations in the observer. Cognition, 125, 26–36. Ticini, L. F., Rachman, L., Pelletier, J., & Dubal, S. (2014). Enhancing aesthetic appreciation by priming canvases with actions that match the artist’s painting style. Frontiers in Human Neuroscience, 8, 391.