2288 IEEE TRANSACTIONS ON CYBERNETICS, VOL. 44, NO. 12, DECEMBER 2014 Expressive Body Movement Responses to Music Are Coherent, Consistent, and Low Dimensional Denis Amelynck, Member, IEEE, Pieter-Jan Maes, Jean Pierre Martens, Senior Member, IEEE, and Marc Leman Abstract—Embodied music cognition stresses the role of the human body as mediator for the encoding and decoding of musical expression. In this paper, we set up a low dimensional functional model that accounts for 70% of the variability in the expressive body movement responses to music. With the functional principal component analysis, we modeled individual body movements as a linear combination of a group average and a number of eigenfunctions. The group average and the eigenfunctions are common to all subjects and make up what we call the commonalities. An individual performance is then characterized by a set of scores (the individualities), one score per eigenfunction. The model is based on experimental data which finds high levels of coherence/consistency between participants when grouped according to musical education. This shows an ontogenetic effect. Participants without formal musical education focus on the torso for the expression of basic musical structure (tempo). Musically trained participants decode additional struc- tural elements in the music and focus on body parts having more degrees of freedom (such as the hands). Our results confirm earlier studies that different body parts move differently along with the music. Index Terms—Dynamical systems, embodied music cognition, functional data analysis, musical expressiveness. I. I NTRODUCTION T HE POWER of music as a nonverbal expressive com- munication system is widely recognized [1]–[4]. Yet, the mechanisms that support the encoding and decoding of musical expression are still poorly understood, especially in social contexts (e.g., pop concerts, joint action, etc.). Embo- died approaches to music have defined the human body and body movements as core aspects of these encoding-decoding mechanisms [1], [5]. In general, body movements are consid- ered to facilitate the nonverbal expression and communication of emotions, feelings, ideas, and intentions [6]. Manuscript received February 20, 2013; revised October 15, 2013; accepted January 29, 2014. Date of publication February 28, 2014; date of current version November 13, 2014. This work was carried out in context of the EmcoMetecca Methusalem Project at Ghent University, and suported by the Flemish Government. This paper was recommended by Associate Editor M. M. Carvalho. D. Amelynck and M. Leman are with the Department of Arts and Philosophy, Ghent University, Gent 9000, Belgium (e-mail: denis.amelynck@ugent.be; marc.leman@ugent.be). P.-J. Maes was with the Department of Music and Psychology, McGill University, Montreal, QC H3A 0G4, Canada, and is now with the Department of Arts and Philosophy, Ghent University, Gent 9000, Belgium (e-mail: maes.pieterjan@gmail.com). J. P. Martens is with the Department of Engineering of the Ghent University, Gent 9000, Belgium (e-mail: jeanpierre.martens@ugent.be). Digital Object Identifier 10.1109/TCYB.2014.2305998 In the context of music production, expressive movements can be encoded into sound (e.g. [5], [7]–[15]), typically through the use of a musical instrument. Accordingly, the structural features inherent to a musical composition (e.g., melodic lines, rhythm, etc.) combined with the expressive performance of a musician (e.g., timing, dynamics, etc.) cre- ate, what has been called, moving sonic forms [1]. When listening to music, people can mirror the expressive aspects of moving sonic forms back into actual movement patterns. Synchronization of movement to the musical beat is known to be based on brain regions that associate sounds with motor activity [16]. However, people are also capable of generating smoother body movement patterns that go along with the musical expression [15], [17]–[19]. These movement patterns can be further connected to other modes with which actions are typically associated, like emotions, situations, and images. By mirroring sound to movement, music can be experienced and understood as intentional, expressive, and semantically meaningful [1]. A successful and effective communication of musical ex- pression requires that human expressive movement responses to music are at least partly coherent and consistent. There- fore, the study of patterns of coherency and consistency in music-evoked body movements is important to provide deeper insights into musical signification processes in a social con- text. Sufficient social alignment is required before individual expressiveness can be assessed [20]. Accordingly, we want to be able to define what is common in the expressive response of a population (commonality), as well as what is different in the expressive responses of the individuals of this population (individuality). So far, only a small number of studies have addressed coherence/consistency in expressive movement responses to music. Leman et al.[18] used a regression model to study the coherence of listeners’ movements in response to Guqin music. It was shown that there was a trend depending on learning. Desmet et al.[21] applied dynamic time warping (DTW) and cross correlation to estimate group coherence of spontaneous movements to music. Based on recurrence analysis, Varni et al.[22] calculated the level of synchro- nization established among the affective behaviors of each single subject in the group on the basis of a generalized autocorrelation function. Several other studies have tried to capture the coherence/consistency of group movements [23]. However, in several of the recent studies that consider music- related experiments in relation to expressiveness [9], [24], 2168-2267 c  2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission. See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.