MUSIC AND MIND -- A THEORY OF AESTHETIC DYNAMICS* Brian D. Josephson\a and Tethys Carpenter\b Note: this material is subject to copyright restrictions and may not be used for for any commercial purpose without the permission of the copyright owner. If it is transmitted, copied or printed, then this notice, any other information herein relating to copyright, and author and publisher names, must be included in the derivative material (except for the case of brief quotes for the purposes of review or research). ABSTRACT It is argued that purely perceptual or generative accounts of music are inadequate to account for its specificity, and that proper accounts of music must take into account also a more fundamental level of the mind (or of consciousness), a level we term the 'aesthetic subsystem'. The latter constitutes a domain of universality and of meaning that acts in conjunction with more peripheral aspects of cognition. Suggestive parallels with systems such as biosystems and lasers are used to account for a number of features of musical processes in terms of the model. *Paper submitted for the Proceedings of the Conference 'Selforganisation as a Paradigm in Science' held in Kaiserslautern, Germany, September 28-30, 1992. [published by Springer, 1994, under the title 'On Self-Organisation', ed. Mishra, Maass and Zwierlen]. ((c) Springer-Verlag) a\Cavendish Laboratory, Madingley Road, Cambridge CB3 0HE, U.K. b\Department of Music, Royal Holloway and Bedford New College, Egham, Surrey TW2 0EX, U.K. * * * * * * * * MUSIC AND MIND -- A THEORY OF AESTHETIC DYNAMICS Brian D. Josephson and Tethys Carpenter 1. Introduction: The problem of music and meaning The question 'what is music?' is a subtle one, which has been approached from a number of different points of view by different authors. Lerdahl and Jackendoff (1), for example, on the basis of examination of the structure of musical compositions, propose a number of 'generative principles', which they hypothesise can serve to distinguish actual music from arbitrary sequences of musical elements; while Bregman (2) focuses his attention on the perceptual processes involved in musical perception, arguing that the latter can be expected to utilise the same cognitive mechanisms as those involved in ordinary auditory perception. The skill of the musician, according to this point of view, is in essence the skill of assembling musical elements in such a way that the mind will put them together to form a complex 'auditory scene', analogous to an ordinary visual scene. A third approach is that of Suzanne Langer (3), who regards a piece of music as something functioning as a symbol, and which can thereby create in our minds ideas related to feelings. This symbol aspect, implying a distinction between what we perceive as happening in the music (the statement and development of themes, for example), and the effects that these percepts have on us, is not explicitly addressed in either of the approaches previously mentioned. The concept that there is an 'idea' and not just a structure or a scene can be illustrated by an example, such as that of a person listening to the second movement of Mozart's piano concerto no. 21 in C major. A few bars of this movement are sufficient to create a very specific mood, which is sustained and amplified through the whole of the movement. Listening to it being played, the dominant impression is of how well the musical choices made by the composer fit into the whole situation in this regard, the choices made having the appearance of being both appropriate and near optimal. These features apply equally well to music of a more intellectual character. Such phenomena in music fit in with Suzanne Langer's ideas (particularly since one of the characteristic features of a symbolism is its specificity), but tend to suggest that there are more constraints on good music than can be accounted for in terms of approaches such as the two first mentioned. Again, music generated by applying such rules as have been proposed, while in some sense sounding like music lacks, to an experienced listener, the quality of real music. We are faced, then, with the existence of a phenomenon that is apparently at once very specific and very arbitrary (the latter in the sense of there being no clear reason for the observed specificity). Langer's ideas, being essentially statements of listeners' intuitions, have little to say about underlying mechanisms. As a step towards the formulation of hypotheses concerning the latter, we take our cue from situations in physics that are analogous, such as the existence of highly specific and reproducible atomic spectra. The latter, while observable using very simple equipment operating on a macroscopic scale, can be comprehended only in terms of the physics of a different domain, the domain of the atom as understood in the light of the quantum theory. We seem virtually forced to view music in a similar fashion, that is to say as the outward expression of more fundamental phenomena occurring at deep levels of the mind, or of consciousness. This paper is devoted to the exploration of the feasibility and utility of such a concept. The basis of the present formulation is the postulate of the existence of an aesthetic subsystem, which supports specific aesthetic processes, and is presumed to be distinct from both perceptual processes and processes of rational analysis (in which respect it should be noted that while we can analyse an art object with a view to understanding what factors are involved in our appreciation of that object, such rational analysis is not a precondition for such aesthetic appreciation to occur). Perception of music is not 'mere perception' but perception allied to the presence of a different, more fundamental system. In subsequent sections of this paper, this idea will be developed in some detail. But before doing this, we note the existence of a close parallel with natural language, in that comprehension of natural language involves more than mere perception of its structure: knowledge of meaning and knowledge of the corresponding object domain are involved also. We do not, in view of the subtleties of the constraints involved in both cases, consider that these additional factors can be adequately taken into account by means of a closed set of semantic 'generative rules' as Lerdahl and Jackendoff (1) seem to believe, either in the case of language or in the case of music. Some comment should be made finally on the work of Cooke (4), who attempted, in a somewhat simplistic fashion, to develop the idea that music is a specific language whose elements could be discovered by understanding their roles in particular pieces of music. This idea that music is a language has been challenged by Langer (3), although certain aspects of Cooke's ideas are consistent with the proposals developed here. 2. Details of the aesthetic process 2.1 As already indicated, the aesthetic process is conceived of as a special aspect of mind that deals with aesthetic matters (in an analogous way to the way that processes of the visual system deal with vision). We can be consciously aware of the outputs generated by the aesthetic processes, just as we can be consciously aware of the visual percepts that the processes of visual perception generate. 2.2 Activity of the aesthetic subsystem can be induced in various ways. One is the process of listening to music, and another the process of composition. These differ, obviously, in that in the former case the stimulus comes from outside, while in the latter case the aesthetic state is self-induced (while performance of music involves a combination of both features). 2.3 By virtue of the fact that music can be considered as information, it follows that information plays a significant role in the functioning of the aesthetic subsystem. In listening, information fed in determines the state of the aesthetic subsystem. Conversely, in composition the aesthetic subsystem generates information (see, however, sec. 3.2). The importance of information is one of a number of aspects in which aesthetic processes parallel life, where information (e.g. DNA) plays a similarly important role (parallels with life have been noted from a different point of view by Langer (5) and Schoenberg (6). 2.4 We are concerned therefore with a two-way process, where information affects state and state generates information (for example, a particular rhythm in conjunction with other aspects of the music is a generator of a particular mental state (state of the aesthetic subsystem) and a mental state may tend to generate a particular rhythm. This leads one to hypothesise that self-sustaining loops may play an important role in aesthetics (see the subsequent discussion of composition in Sec. 3.2). A similar feature is found in life, where enzymes may catalyse particular subprocesses which in turn regenerate the same enzymes. According to this picture, the rhythms or other features (e.g. harmonies or melodic patterns) that tend to be adopted for music are the ones that tend to be self-sustaining in the manner indicated. 2.5 This picture can now be broadened to the more comprehensive one indicated in fig. 1. This assumes that a variety of phenomena found in music are directly connected with unmanifest processes occurring at the deeper aesthetic level (in the same way that atomic spectra and the Raman effect are visible consequences of corresponding phenomena at the atomic level). We hypothesise, for example, that particular themes that are perceived in the music, and which are seen to develop in particular ways, are connected with particular subprocesses and their development at the aesthetic level. These subprocesses in the aesthetic system have specific effects on each other (parallelling analogous behaviour characteristic of biosystems), in a way that has visible consequences in the way themes or musical elements influence each other. 2.6 A number of parallels between aesthetic processes and life processes have already been noted. An interesting further parallel is one between the process of tension creation and tension reduction, in the case of music, and the process of homeostasis in biology. Some aspects of biosystems can be best understood in terms of the existence of prescribed norms and the processes of homeostasis which seek to reestablish these norms when sufficient deviations from them occur. The balance- unbalance condition in biosystems has its musical parallel in tension and its absence, and the specificity of the processes by which biosystems restore balance has its parallel in the specificity of the ways in which musical tensions are resolved. 3. Global level features of music The previous section has been concerned mainly with local details of the structure of music. In this section we step back and examine more global features, rephrasing and extending discussions given elsewhere (7, 8) that were centred around the concept of musical idea. In the current picture music can be defined, roughly speaking, as activity in the aesthetic system which is, to a first approximation, self- sustaining under the given conditions. This condition applies both to a listener who is following the music and appreciating it, and to a composer whose compositional processes are flowing adequately. Thus the listener and the composer have related goals, connected with maintaining the activity of the aesthetic subsystem in an appropriate form. 3.1 The listening process For the listener, the aesthetic system activity is the response to the structured perception of the sound, and he must adequately perceive its structural basis in order to create the full aesthetic response. Familiarity with the style of the music plays a significant role here in making it easier to discover the appropriate structure, while being habituated to a different musical culture on the other hand may have negative effects, in that it may mislead the listener into perceiving incorrect structures, that is to say into perceiving overall structure that does not activate the aesthetic subsystem in the way that it should. But in the long term, that is to say on repeated listening, feedback in terms of the degree of aesthetic response to the presumed structure may enable the listener to adjust his perceptions in the right direction, and thus in the end lead him to hear the music appropriately. 3.2 The process of composition The composer's task is the reverse of the listener's. The principles involved here are considerably more subtle, involving a delicate two-way interplay between the aesthetic subsystem and peripheral systems that link with it. A physics analogy in the form of an analogy with a laser offers useful insights here (see Fig. 2). The characteristic patterns of music have been assumed to be reflections of the underlying dynamics of the aesthetic subsystem. Listening to music involves the response of the aesthetic subsystem to the music, a process parallel to the resonant response of an atom to electromagnetic radiation. If we feed a system of atoms with energy in an appropriate way ('pumping' the system), a process of stimulated emission occurs whereby the atoms emit more radiation than they take in by absorption, a state of affairs which tends to result in the system becoming unstable and generating radiation spontaneously at a frequency close to the frequency of the spectral line. But the precise details of the radiation, e.g. in terms of frequency and spatial distribution, may be strongly influenced by the macroscopic environment of the atomic systems as well; for example, a laser cavity consisting of two parallel mirrors may have a series of resonant modes, and if these are more precisely defined in frequency than is the relevant atomic transition then the laser will oscillate in one of these cavity modes. By analogy, we expect that in the musical case, under appropriate conditions, the aesthetic subsystem activity will take a form determined both by what is possible in terms of the aesthetic subsystem dynamics (corresponding to the atomic resonance) and the composer's learned musical repertoires (corresponding to the modes of the cavity). The composer's ability to create music depends on the appropriate musical environment existing in his peripheral systems, just as the physicist's ability to create laser radiation depends on his ability to provide an appropriate electromagnetic environment for the laser radiation; in both cases, adequate matching for the two subsystems must exist. 3.3 Time dilation Some composers (e.g. Mozart) have claimed to have seen a whole new composition in a flash. This facility poses no particular problem for the present point of view, provided that the properties of the aesthetic subsystem and the linked peripheral systems are such that it is possible to accelerate their activities without any significant change in form. 4. Meaning in music The problem of the meaning of music is a very subtle one. Meyer (9) distinguishes between internal meaning or reference in music, and external reference. If music had only internal reference, listening to music would be a very introverted kind of activity. However, psychological experiments designed to discover specific external references, i.e. links between specific pieces of music and specific types of events, have been generally unsuccessful. There are various possible reasons for such a failure, one of them being that the abstract situation, where a subject is asked whether a piece of music A reminds him of situation B, may be significantly different operationally from the corresponding concrete situation of placing a subject actually in situation B, and asking him how he feels about music A then. Meyer's suggestion (9) that the reference of music may be archetypical in nature (e.g. the whole complex of feelings associated with death) rather than conceptual may also be relevant. Let us now see what the picture presented here has to say about such matters. One way to describe meaning is to say that it is a way of talking about certain kinds of cause and effect in the context of information processing (10). From this point of view, the initial meaning of a composition may be considered to be the corresponding activity in the aesthetic subsystem. Component parts of a composition equally have meaning in respect of the various specific effects that they induce in the aesthetic subsystem. The 'meanings' considered thus far have been in the category defined by Meyer as inner meaning. As already noted, the question of external meaning or reference is unclear. However, the problem can be approached via the fact that music appears to have the capacity to enhance other kinds of activities (for example, ballet and opera). Sometimes, for example, an opera may have an inadequate libretto or be performed or acted badly, but the performance remains effective because of the music. It might be said in interpretation of this fact that the music adds information that is missing from the performance (in the same way that a commentary or a caption adds missing information to a picture). The aesthetic subsystem can thus be considered to have an informative function, perhaps closely connected with intuition or understanding in general. 5. Parallels between music and exposition The use of phraseology such as 'the statement of a theme' or 'the development of a theme' points to the existence of parallels between music and the exposition of a collection of ideas. Such parallels are to be expected in a model where informational input is building up complex structures, and it may be anticipated that similar universal principles or processes might be involved in both situations. One such principle is the principle of reference, i.e. the connections that exist between simple surface forms (e.g. words in the case of language; particular patterns of notes in the case of music) and deeper entities. In the case of language, words are used to gain access to particular pieces of knowledge or activities based on that knowledge. In the case of music, patterns of notes provide access to musical processes (cf. Fig. 1), for example starting a process, developing it or bringing it to a close. However, music and natural language are in many ways not at all equivalent. For example, with language, meanings (which are mechanisms whereby different processes may be linked) are learnt associations, whilst in the musical case the dynamical factors linking different processes in the aesthetic subsystem are assumed to be universal. This is not, however, to say that learning is unimportant or irrelevant in the musical case. But in the musical case learning, as has been explained above, is taken to be operative at more peripheral levels (e.g. the level of auditory perception). 6. Concluding comments We started off arguing that a conventional perceptual account of music was inadequate, and that something corresponding to the subtle influence of the products of the perceptual process on the mind ought to be included in cognitive models as well. A model incorporating an 'aesthetic subsystem' was developed, and used to give an account of a number of processes associated with music. The essence of the model is that aesthetic processes have a certain universality, making them essentially independent of the ordinary biological domain but perhaps akin to mathematical intuition. While these are somewhat unusual hypotheses, it is hard to see how the specificity of music can be accounted for in any other way. For the future development of the model, one possible avenue is the question of to what extent very specific regularities of music discussed by investigators such as Lerdahl and Jackendoff (1) and Narmour (11) can be accounted for in a transparent manner on the basis of the model. Such connections might enable one to characterise specific processes in the aesthetic subsystem, and to understand their cognitive function, perhaps producing ultimately accounts of the workings of the aesthetic (or intuitive) subsystem of a similar kind to those that we now have of life. Finally, attempts to account for the semantic aspects of music in detail, such as those mentioned in the introduction due to Cooke (4), can perhaps be usefully extended on the basis of concepts such as those developed here. Acknowledgement We are grateful to Dr. Ian Cross for helpful comments on drafts of the manuscript. References and footnotes 1. F. Lerdahl, R. Jackendoff: A Generative Theory of Tonal Music (MIT Press, London, Cambridge, Mass. 1983). 2. A. S. Bregman, Auditory Scene Analysis: the Perceptual Organization of Sound (MIT Press, 1990). 3. S. K. Langer, Philosophy in a New Key (Harvard University Press, London, Cambridge, Mass. 1951). 4. D. Cooke, The Language of Music (Oxford University Press, Oxford, New York 1989. 5. S. K. Langer, Mind: an Essay on Human Feeling (Abridged edn.: Johns Hopkins University Press, Baltimore 1988, chap. 7). 6. A. Schoenberg, Folkloristic Symphonies, in Style and Idea: Selected Writings of Arnold Schoenberg (ed. Leonard Stein, Faber and Faber, London 1975). 7. B. D. Josephson and T. L. Carpenter, New Scientist 129(1762), 2 (1991). 8. B. Josephson and T. Carpenter, New Scientist 131(1780), 51-2 (1991). 9. L. B. Meyer, Emotion and Meaning in Music (Univ. of Chicago Press, Chicago 1959). 10. D. J. Bohm, Meaning and Information in The Search for Meaning (ed. P. Pylkkaenen, Crucible, Wellingborough, Northants., U.K. 1989) p. 43 P 62. 11. E. Narmour, The Analysis and Cognition of Basic Melodic Structures: the Implication-Realization model (University of Chicago Press, Chicago 1990). Figure Captions 1. Illustrating the way musical phenomena in perception (or in the imagination), such as A and B in the diagram are linked with deeper phenomena such as a and b in the postulated aesthetic subsystem. The interactions and dynamics in the aesthetic subsystem have a strong influence on the observed structure of music, in the same way that the dynamics of the atom influences the observed behaviour of atomic spectra. 2. The composer's aesthetic system and the peripheral systems mutually influence each other, in the same way that in the laser the atoms and the radiation field mutually influence each other. In either case, the two subsystems in isolation have their own characteristic modes of behaviour, and when the subsystems are coupled together the observed behaviour has features deriving from the modes of behaviour of the isolated subsystems (see sec. 3.2 for a fuller discussion). (N.B. pictures omitted in this ASCII file -- they convey no information that a basic exercise of the imagination using the above will not do).