The Embodied Experience of Abstract Art: An Exploratory Study

Abstract Postural control of people looking at 10 paintings of distinct artistic styles (5 Mondriaan’s neoplasticism and 5 Pollock’s action painting; N = 30) was investigated, using recurrence quantification analysis. Also, we explored the relation between postural control measures, emotional states of being moved (“Moved-By” and “Drawn-Towards”), and aesthetic appraisal (“Beauty” and “Complexity”). Results revealed differences in postural control, most notably more dynamically stable and complex postural sway in both the medial–lateral and anterior–posterior direction for Pollock paintings. Emotional states of being moved and aesthetic appraisal were associated with postural control across paintings, in several ways. For instance, paintings that people found more beautiful and felt more drawn towards, on average also tended to make them lean forwards more. Together these findings provide preliminary empirical evidence for the embodied nature of art experiences.


Introduction
Art is well-known for its ability to move people, both literally and figuratively, physically and emotionally.However, art as an embodied activity, fusing in a way the artwork (as artefact with various spatial and possibly temporal properties), the viewer's movements (i.e.actual bodily motion) and emotions (like being moved and aesthetic evaluations such as of beauty), and numerous other contextual aspects, is relatively understudied.Contemporary theories about art experience have tried to address the importance of the body and of bodily movement in various ways, by observing increased activity in brain regions such as the motor control systems (e.g.Chatterjee & Vartanian, 2014) and applying simulation theory (e.g.Freedberg & Gallese, 2007), by assessing people's perceptions of movement in terms of emotional states of being moved (e.g.Gerger et al., 2018) and bodily sensations (e.g.Schino et al., 2021), by considering their actual bodily movements of various kinds (e.g.K€ uhnapfel et al., 2024), and by pointing at the role of motion features in the artworks themselves (e.g.Darda et al., 2022;Van Geert & Wagemans, 2020).The present study explores the triangular relation between art, body and mind by investigating people's experience of abstract paintings in an empirical way from a radical embodied perspective (e.g.Chemero, 2009).Angelino (2015) argues that movement resides in shapes which are created by the artist, and which, although immobile in reality, create a dynamical (dis)balance in the artwork.In turn, this gives rise to distinct bodily movements in the viewer.That is to say, the viewer is literally moved by a visual work of art.Brinck (2018) follows that bodily movements in the art experience are impelled by the viewer and the artwork conjointly.Together the artwork, with its dynamic balance, and the viewer, as an active, embodied experiencing organism, form a temporary coupled system.It creates a link between productive (e.g.painting the artwork) and receptive (e.g.looking at the painting) experiences of the artist and the viewer, connecting their movements over space and time by the attunement to the artwork (van Klaveren et al., 2019).In this view, art experiences are emergent phenomena resulting from people's attunement to art (e.g.Cox et al., 2023).Earlier studies have reported how artworks are able to elicit distinct movement patterns, such as postural sway and eye movements in viewers (e.g.Ganczarek et al., 2015), both of which are indicators of people's attunement to the environment.These studies showed that viewers' postural sway is influenced by looking at visual artworks such as paintings and sculptures.It has been proposed that pictorial depth and (illusory) motion present in paintings lead to more postural sway (Kapoula et al., 2011(Kapoula et al., , 2015;;Kapoula & Gaertner, 2015).This could be directly caused by such 'visual cues', but (changes in) fixations and also mental imagery has been proposed to play a role (Ganczarek et al., 2015).Moreover, Lelard and colleagues (2019) suggest that dynamic postural patterns in response to visual stimuli may reflect attentional effects of emotional states (cf.Pellecchia, 2003;Ramenzoni et al., 2007).Fuchs and Koch (2014) argue that these bodily movements go hand in hand with emotional states of being moved and aesthetic appraisals such as of beauty.For instance, Gerger, Pelowski and Leder (2018) show that intense bodily reactions are related to increased emotional intensity, higher feelings of being moved and more positive evaluations such as liking.According to Menninghaus et al. (2015), feelings of being moved comprise high-intensity emotional responses combining positive and negative emotions such as joy and sadness while involving only low-to-mid self-reported levels of arousal.Moreover, actual bodily movements such as laughter, tears or chills and action tendencies such as approach and avoidance behaviours seem to be closely related to aesthetic appraisal in terms of pleasantness, familiarity or novelty (Menninghaus et al., 2019).In particular, unpleasant images seem to affect body posture in terms of avoidance or freezing responses (Lelard et al., 2019).While the perception of being moved may be an emotional response to a work of art, directly resonating with the viewer, liking along with beauty and goodness may be described as emotional evaluations of an artwork's aesthetic virtues, assessing its hedonic qualities (Pelowski et al., 2016).Hur et al. (2020) suggest that beauty is related to high levels of happiness, and low levels of fear and arousal in contrast to sublimity which is linked to the reverse pattern based on selfreports.Moreover, Sarasso et al. (2020) propose that beauty pertains to motor inhibition following the "stopping for knowledge hypothesis."At the same time, more dynamic and approaching bodily movements are frequently associated with increased interest and liking (K€ uhnapfel et al., 2024).However, the relationships between observed movement patterns and subjective measures have not been thoroughly investigated from a radical embodied perspective yet (Cox et al., 2023).In particular, it has not been examined whether and how postural control reflects people's differential attunement to different abstract painting styles, and their influence on emotional states and aesthetic appraisal.

The present study
The present study explored the embodied nature of art experiences in terms of the postural control of participants looking at collections of abstract paintings.More specifically, we compared a selection of hallmark paintings by Piet Mondriaan and Jackson Pollock.The artistic styles of these painters undeniably differ in many respects.First and foremost, Pollock's action-art compositions are considerably more spatially complex than Mondriaan's creations consisting of basic geometrical shapes (e.g.Locher et al., 2005;Taylor, 2002).In addition, Mondriaan's paintings are considered "low motion," whereas in the same account, Pollock's paintings are labelled as being "high motion" (Humphries et al., 2021).Both leaning posture (Chisholm et al., 2013) and postural sway will be analysed, as part of postural control.Besides measures of the amount of sway, we also quantified the dynamic organisation of postural control using recurrence quantification analysis (RQA).It can be expected that postural control while looking at compositions of low spatial complexity and low motion is itself lower in complexity and motion, than while looking at compositions of high spatial complexity and high motion (Kapoula & Gaertner, 2015).Specifically, we predicted postural sway to be more dynamically complex and variable for the Pollock paintings.In addition, we measured participants' feelings of being moved-by and drawn-towards the paintings to assess their emotional states of being moved, as well as their aesthetic appraisal of the paintings in terms of beauty and complexity.We expected that engaging with a work of art gives rise to specific relationships between their patterns of postural control, emotional states of being moved, and aesthetic appraisals.We figured that the expressive spatial configurations of abstract paintings, in particular, would provoke considerable differences in people's emotional states of being moved and aesthetic appraisals.By extension, paintings of markedly different artistic styles, could amplify these effects, as well as create the clearest associations between postural-control measures and emotional states as well as aesthetic appraisals.Two specific predictions following from this are that paintings' subjective spatial complexity and the dynamic complexity of the postural control while looking at them would be positively correlated, and similarly for the emotional state of being moved and the postural-sway variability across paintings.

Participants
In two separate rounds of data collection, 35 undergraduate students of the University of Groningen participated in this study either in exchange for course credits or without reward.All participants had normal or corrected-to-normal vision, and none of them reported colour blindness.Five participants were discarded from further analyses because of equipment malfunction or dizziness during the experiment.The 30 resulting participants were 14 females and 16 males, aged 18-29 years.The study was approved by the local ethics committee of the Department of Psychology at the University of Groningen, with approval number PSY-2122-S-0093.Written informed consent was retrieved from all participants.

Materials and procedure
While standing without shoes on a Wii Balance Board (WBB; Nintendo Company, Ltd.) at a distance of 1.9 m, participants looked for 45 s at 10 life-sized pictures of paintings in a random order presented on a large monitor (178 � 99 cm; 1920 � 1080 pixels; 60 Hz; Samsung Group).Five paintings were by Jackson Pollock and five by Piet Mondriaan (Table 1).The paintings were presented using OpenSesame (version 3.3) using the backend psycho.This was synchronised with a custom-built software application, communicating with the WBB through a Bluetooth wireless connection, which registered postural-sway data at a fixed sample rate of 100 Hz.Before the experiment started, calibration of the WBB was performed.The WBB has been proven to be a less expensive, but valid and reliable alternative to the "regular" laboratory force plates (e.g.Clark et al., 2018;Pavan et al., 2015).After its presentation, the painting disappeared from the screen and the postural sway measurement ended.
Participants were then handed a tablet (iPad; Apple, Inc.) running Qualtrics, on which they answered four questions in a self-paced manner.The painting was made visible again during this phase.Items inquired, on a scale from 1 to 10, about participants' emotional state of being moved in terms of being "Moved-By" ("How much are you emotionally moved by the painting that you have just seen?") and being "Drawn-Towards" ("How much are you emotionally drawn towards the painting that you have just seen?"), as well as about the participants' aesthetic appreciation of the painting in terms of "Beauty" ("In your opinion, how beautiful is this painting?")and "Complexity" ("In your opinion, how complex is this painting?").After the tablet was retrieved from the participant, the next painting was presented.The entire experiment lasted about half an hour, including (optional) break and debriefing.

Data analysis
Postural control was analysed in terms of the centre-of-pressure (COP) movement in the medial-lateral (ML; side-to-side) direction and anterior-posterior (AP; back-andforth) direction.Movement analyses were performed on 30 s of measurement (i.e.3000 data points), after low-pass Butterworth filtering.In accordance with the explorative nature of this study, several COP measures were considered.Leaning posture (i.e. the overall tendency of leaning left/right and towards/away) was operationalized, as the average of the COP movements in each direction (AV ML and AV AP ).The spatial magnitude of the postural-sway variability was operationalised as the standard deviation of the COP movements in each direction (SD ML and SD AP ).The dynamic organisation of the postural control was analysed by applying recurrence quantification analysis (RQA) on the COP movements in each direction separately.
RQA is a nonlinear time-series technique which combines phase-space reconstruction using a delay embedding technique (Takens, 1981) with the quantification of the recurrence plot (RP; Eckmann et al., 1987).RPs are a tool to visualise the behaviour of dynamical systems based on recurrences (i.e.revisited points) in their trajectories in reconstructed phase space.From the RP, the dynamical organisation of the system can objectively be quantified by means of several recurrence measures.The analysis requires certain parameters to be set, related to the phase-space reconstruction and to the quantification of recurrent patterns.For the first, the delay and the embedding dimension needed to be chosen.Here, we used a delay of 70 and an embedding dimension of 6, using the Average Mutual Information (Fraser & Swinney, 1986) and False Nearest Neighbour (Kennel et al., 1992) procedures, respectively.For the second, we used a variable radius in phase space, which provided recurrence rates of 0.10 for each trial (cf.Riley et al., 1999;Wijnants et al., 2009).Finally, the minimum length of the diagonal and vertical lines was set to 20 so as to avoid relatively short but random diagonals leading to saturation in the line measures (Almeida et al., 2018, Thiel et al., 2002;Tommasini et al., 2022), and leading to the identification of recurrent patterns of at least 200 ms long.Recurrent patterns indicate deterministic episodes of various duration in the dynamics of the underlying system, which appear as diagonal and vertical lines in the RP.
The next step in the analysis is to quantify the recurrent patterns in the RP.We considered five recurrence measures, for each COP movement direction: Determinism, Meanline, Entropy, Laminarity, and Trapping Time.Determinism (DET ML and DET AP ) is defined as the ratio of the number of recurrent points forming recurrent patterns on diagonal lines of minimally 20 recurrent points over the total number of recurrent points.This measure informs about the extent of deterministic behaviour of the system.Meanline (MNL ML and MNL AP ) is the average duration of diagonal recurrent patterns during the measurement, that is, the average length of diagonal lines.It is considered a measure of dynamic stability of the system.Entropy (ENT ML and ENT AP ) is computed as the Shannon entropy of the distribution of line lengths of the diagonal lines in the RP (cf.Webber & Zbilut, 1994).Entropy quantifies the dynamic complexity of the system.Laminarity (LAM ML and LAM AP ) and Trapping Time (TT ML and TT AP ) capture the proportion and average duration of laminar states of minimally 200 ms, respectively.Laminar states represent intermittency or rigidity in the system, that is, when the system gets stuck in one or more states for some time.Laminarity is calculated similarly as Determinism except that it reflects the proportion of recurrent points on vertical lines in the RP.Trapping Time is similar to Meanline but quantifies the average vertical line length, and therefore the average duration of vertical recurrent patterns during the measurement.Detailed descriptions of RQA including descriptions of all measures can be found elsewhere (e.g.Marwan et al., 2007;Riley et al., 1999;Wallot & Leonardi, 2018;Webber & Zbilut, 2005;Wijnants et al., 2012).RQA was performed using custom-made Matlab scripts in combination with a function from Marwan's CRP Toolbox (http://www.recurrence-plot.tk).
Because most of the COP measures were not normally distributed, non-parametric tests were used to check our expectations.For comparing the two artistic styles Wilcoxon signed-rank test with a Monte-Carlo resampling procedure (10,000 samples) were used on the seven COP measures in both movement directions (one-tailed), as well as on the ratings of the four questions (two-tailed).In addition, Spearman's rank correlations (q) across paintings were calculated between the over-participants-averaged COP measures and ratings.For this, the average over all the participants was calculated of each measure and rating, for each of the paintings separately.These were subsequently correlated across paintings.This type of analysis gives insight into how bodily movement, on the one hand, and emotional states and appreciations, on the other hand, are associated (across paintings rather than across people).With this, the paintings are essentially regarded as individual probes for such associations underlying art experiences.Extreme outliers (i.e.below −3 SD and above þ3 SD) with respect to COP total path length were removed.For all analyses a was set to .05, and in line with the exploratory nature of this study no corrections for multiple comparisons were made (see Armstrong, 2014).Statistical analyses were done in SPSS.

Results
Table 2 provides an overview of all the COP measures and the ratings to the questions for each of the ten paintings.Largely in line with our main expectation, looking at Pollock paintings went along with more deterministic, dynamically stable and complex postural sway, which consisted of longer intermittent episodes in the dynamics.The significant difference AV AP , Z ¼ −2.18, p ¼ .013,r ¼ .28,indicated that there was a tendency to look at Pollock paintings with a slightly more forward leaning posture.
With respect to the ratings, Wilcoxon tests yielded significant differences between the two artistic styles for all four questions: Beauty, Z ¼ −1.94, p ¼ .027,r ¼ .25,Complexity, Z ¼ −2.27, p ¼ .011,r ¼ .29,Moved-By, Z ¼ −1.81, p ¼ .034,r ¼ .23,and Drawn-Towards, Z ¼ −1.75, p ¼ .041,r ¼ .23.This means that Pollocks were regarded as more beautiful and more complex than Mondriaans, and with respect to their emotional states of being moved, participants' felt more moved-by and more drawn-towards Pollock paintings.

Across-painting propensities of art experience
In line with our second expectation, emotional states of being moved and aesthetic appreciations across the ten paintings, averaged over all participants, were associated to the averages of the COP measures in several ways.Paintings which rated higher in Complexity elicited more deterministic and laminar postural sway in the medial-lateral direction, q(10) ¼ .71,p ¼ .022(DET ML ), and q(10) ¼ .64,p ¼ .044(LAM ML ), respectively.Painting complexity was also strongly associated with the dynamic stability and complexity and average length of intermittent episodes in the anterior-posterior postural sway, q(10) ¼ .87,p ¼ .001(MNL AP ), q(10) ¼ .86,p ¼ .002(ENT AP ), and q(10) ¼ .92,p < .001(TT AP ), respectively.Paintings were leaned towards more when they were rated higher on Beauty, q(10) ¼ −.66, p ¼ .038,and on being Drawn-Towards, q(10) ¼ −.90, p < .001.For paintings that participants were more Moved-By on average, also more dynamically stable and dynamically complex postural sway in the anterior-posterior direction was observed, q(10) ¼ .87,p ¼ .001(MNL AP ), and q(10) ¼ .79,p ¼ .006(ENT AP ), respectively.More emotionally moving paintings also elicited longer laminar anterior-posterior postural-sway patterns, q(10) ¼ .85,p ¼ .002(TT AP ).

Discussion
Overall, the results of this study partially confirm our expectations.People looking at paintings of two markedly different artistic styles displayed postural control which differed significantly in its dynamic organisation.That is, COP measures differ, to some extent at least, and appear to be related to how paintings are experienced.First, the spatial extent of the postural-sway variability, which was expected to be higher for Pollocks than for Mondriaans, did not diverge.This means that the low-motion and high-motion distinction, which has been made between Mondriaans and Pollocks (Humphries et al., 2021), did not directly translate in more or less variable (i.e. less or more stable) postural control.
However, dynamic stability and complexity of the postural sway were found to be higher for Pollock paintings than for Mondriaan paintings.The difference in dynamic complexity could be revealing that differences in spatial complexity between Pollock paintings and Mondriaan paintings are reflected by differences in the temporal complexity of the embodied aesthetic experience.Importantly, at present, it remains speculation whether it is the spatial complexity (or the motion distinction or any other aspect) of the painting that directly leads to the movement patterns.It could merely be a side effect of some other (art-related) perceptual, emotional and cognitive process.To investigate this, additional and more detailed experimental studies are necessary.This would, at least, require using both an objective measure for spatial complexity and a more targeted selection of paintings.
It is well-known that postural control is responsive to environmental circumstances, in order to maintain a dynamically stable posture, while at the same time, it serves an explorative function in actively seeking out information (see e.g.Balasubramaniam & Wing, 2002;Riley & Turvey, 2002).Additionally, posture has been shown to be reflective of cognitive processes (e.g.Balaban et al., 2004;Pellecchia, 2003;Rosenbaum et al., 2017).In this light, the result that dynamic stability and complexity of the postural sway were different between Pollock and Mondriaan paintings is particularly interesting.Although these abstract painting styles differ in various ways, stability and complexity are two of the spatial features on which they deviate quite considerably.Without intending to oversell these particular findings, this does provoke an interpretation in which the dynamical organisation of bodily movements during (abstract) art experience tends to reflect at least some of the compositional characteristics of the artwork.
Some additional clues about the embodied nature of art experiences might partly be provided by the strong correlations, which were found between the subjective appraisals of the paintings and the measures of art-elicited postural control.Aesthetic appreciations in terms of beauty and complexity or their relation (Forsythe et al., 2011) and emotional states of being moved like feeling moved-by or drawn-towards, are by no means pure and direct indicators of perceived spatial structure, artistic quality, or any such thing.Nor do they reveal in a straightforward manner how those things might have influenced bodily movement.Such subjective measures merely provide people with a means to express how they experienced the artwork emotionally, cognitively, bodily and otherwise.However, we know that these things are linked reliably, albeit in nonobvious ways, to various features of artworks (Arnheim, 1974;Van Geert & Wagemans, 2020).In that sense, across-painting correlations like the one found between subjective complexity of the paintings and dynamic complexity of postural sway, provide additional hints at a connection between emotional, cognitive, and bodily aspects of the experience, from a radical embodied point of view.As said earlier, what it is in artworks that creates this link or makes it overt in a specific way, clearly warrants more empirical investigation.
Finally, the differences which were found between the Mondriaans and Pollocks with respect to participants' emotional state of being moved and aesthetic appraisal, should be interpreted with care.The relatively small and homogeneous sample of participants in this experiment implies that these results cannot be generalised to any type of population.Claims about differences in aesthetic appraisal between Mondriaan and Pollock paintings, or between the artistic movements of 'de stijl' and 'action art' for that matter, are unwarranted.It should be stressed though that the goal of this study was not to investigate such things, but to explore the embodied dynamic nature of art experience and discover potential relationships between people's bodily movements while looking at an artwork and their assessment of that piece.For that objective the current sample was quite appropriate, as was the choice for the two different painting styles.

Table 1 .
Title, year of production, and dimensions (cm) of the paintings used.The first five paintings are by Piet Mondriaan; the last five paintings are by Jackson Pollock.

Table 2 .
Averages (over participants) of the COP measures in both movement directions and of the ratings for the four questions, for each of the ten paintings used.(MVD ¼ Moved-By; DRW ¼ Drawn-Towards; BTY ¼ Beauty; CPX ¼ Complexity).