Philosophy of Improvisation

Interdisciplinary Perspectives on Theory and Practice

Susanne Ravn, Simon Høffding, James McGuirk, Susanne Ravn, Simon Høffding, James McGuirk

• Publisher: Routledge
• Year: 2021
• ISBN:
  9781000399141
• Edition: 1
• Topic: Filosofia
• Subtopic: Estetica in filosofia

This volume brings together philosophical and interdisciplinary perspectives on improvisation. The contributions connect the theoretical dimensions of improvisation with different viewpoints on its practice in the arts and the classroom.

The chapters address the phenomenon of improvisation in two related ways. On the one hand, they attend to the lived practices of improvisation both within and without the arts in order to explain the phenomenon. They also extend the scope of improvisational practices to include the role of improvisation in habit and in planned action, at both individual and collective levels. Drawing on recent work done in the philosophy of mind, they address questions such as whether improvisation is a single unified phenomenon or whether it entails different senses that can be discerned theoretically and practically. Finally, they ask after the special kind of improvisational expertise which characterizes musicians, dancers, and other practitioners, an expertise marked by the artist's ability to participate competently in complex situations while deliberately relinquishing control.

Philosophy of Improvisation will appeal to anyone with a strong interest in improvisation, to researchers working in philosophy, aesthetics, and pedagogy as well as practitioners involved in different kinds of music, dance, and theater performances.

Part I
Reconsidering Improvisation

1
The Birth of Planning out of the Spirit of Improvisation

The Iceberg Model

Beth Preston

The formula of our happiness: a Yes, a No, a straight line, a goal.

Friedrich Nietzsche, The Antichrist

Introduction

Is improvisation a qualitatively distinct type of action? If so, it must be distinguished from other, contrasting types of action. With regard to everyday action, especially in philosophy and artificial intelligence, it is contrasted with planning. With regard to the arts, especially music, dance, and theatre, it is contrasted with composition. In both cases, the contrast is between action that is prepared in advance, in some detail, and executed later; and action that is executed on the fly, without such advance preparation.¹ However, this neat contrast falters over the recognition that improvisation shares many important features with planning, including temporally extended structure, orientation towards a goal, and reliance on pre-existing skills, habits and cultural resources (Brown, 2011; Hakli, this volume; Preston, 2013). The contrast thus looks more like a varying, quantitative difference than a sharp, qualitative break.²

This gives impetus to attempts to assimilate improvisation to planning as a mere variant – a move common in artificial intelligence, for instance, where what might otherwise be called improvisation is instead labelled reactive or online planning (Russell & Norvig, 2003; Hakli, this volume). Similarly, improvised music may be assimilated to composition as simultaneous composition and performance (Alperson, 1984). Alternatively – although more rarely – the shared features are taken to license the assimilation of planning to improvisation (Benson, 2003; Hallam & Ingold, 2007). If we want to resist assimilation in either direction – and I think we should – this leaves us with the difficult task of explaining how improvisation and planning are distinct from each other while at the same time intimately related. Some initial attempts characterise them as distinguishable aspects, rather than distinct types, of action (McGuirk, this volume), or as the contrasting endpoints of a continuum (Hamilton, 2000). These are attractive options, but the relationship is still more complex than it appears at first glance, or at least so I shall argue.

In the first section of this chapter, I explain why it is important to resist assimilating improvisation to planning or vice versa. On the one hand, assimilation flies in the face of both the phenomenology of action and the way action is categorised by the behavioural sciences. On the other hand, it represents an implicit evaluation of one type or aspect of action over the other, with implications for social justice issues involving cultural practices. This leaves us with the question of how planning and improvisation are related to each other. In the second section, I offer an “iceberg” model for understanding this relationship. The tip of the iceberg appearing above the water represents the more difficult and therefore noticeable ability to plan, while the much larger part of the iceberg under the water represents the more fundamental but less noticed ability to improvise. I flesh out this model with a detailed discussion of the different ways improvisation and planning make use of the resources available to human agents in an uncertain and constantly changing world. I conclude that we are fundamentally improvisers who occasionally plan, not planners who occasionally improvise.

Against Assimilation

Let us begin with a closer look at a couple of examples of assimilation. It would be difficult to find a clearer case of assimilation of improvisation to planning than the field of artificial intelligence, where action theory is routinely, if implicitly, equated with planning theory. Stuart Russell and Peter Norvig’s standard textbook, Artificial Intelligence: A Modern Approach (2003), defines planning very broadly as “[t]he task of coming up with a sequence of actions that will achieve a goal” (p. 375). They have two chapters on planning. The first covers classical planning – that is, planning in “environments that are fully observable, deterministic, finite, static (change happens only when the agent acts), and discrete (in time, action, objects, and effects)” (p. 375). But the vast majority of real world environments do not satisfy these requirements. So a second chapter covers the planning strategies needed to cope with real world uncertainty. These start with conditional planning, execution monitoring, and replanning, and they end with so-called “continuous planning” which “creates new goals as it goes, and reacts in real time” (p. 455).³ Continuous planning is clearly what we would otherwise call improvisation. In the artificial intelligence context it is seen as merely a variant of classical planning. It is what happens when you loosen classical planning up as far as possible in order to deal with the world as it actually exists for actual agents. The underlying assumption is obviously that continuous planning is not qualitatively different from classical planning but at the opposite end of a quantitative continuum with it.

A clear example of assimilation in the other direction is provided by Bruce Ellis Benson in The Improvisation of Musical Dialogue (2003).⁴ Benson says that improvisation is not a kind of creation ex nihilo – a common misconception – but rather “a kind of reworking of something that already exists” (p. 30). He substantiates this definition with a catalogue of eleven gradated ways this reworking takes place in music, ranging from a performer’s minimal filling in of details not specified in the score to every musician’s situated activity within a musical tradition that they carry forward while modifying (pp. 26–30). So whether there is a definite composition phase preceding a performance phase, as in Western classical music, or whether these phases coincide, as in jazz, musicians “work with the given in order to ‘create’ something new” (p. 32). Benson points out that the gradated nature of his catalogue of types of improvisation shows that the differences between them are quantitative, not qualitative (pp. 30, 146–147). Here, what we would otherwise call executing a plan – that is, performing a composition – is presented as just the most minimal level of improvisation, a mere quantitative variant of the more obviously improvisatory practices in blues or jazz.

In spite of the fact that these two assimilation projects go in opposite directions, they proceed on basically the same two premises. First, they both provide an initial definition of the phenomenon of interest – planning, in Russell and Norvig’s case; improvisation in Benson’s – that is so broad that it cannot very well fail to apply to the whole domain of human action. On the assumption that action is goal-directed, however vague that goal may be, defining planning as coming up with a series of steps to achieve a goal means that action is planned, by default. Whether the series of steps is laid out in advance or constructed on the fly is immaterial. Similarly, on the assumption that action relies on resources of various kinds rather than inventing itself ex nihilo, defining improvisation as the reworking of something already in existence to create something new means that action is improvised, by default. The number and type of resources involved are immaterial. Second, both assimilation projects construe any differences that show up as quantitative, thus derailing any possibility of a qualitative contrast. There is nowhere on the gradient to draw a line and insist on a principled distinction between planning and improvisation. These are the hallmarks of the assimilationist ethos that animates both these views.

Why should we resist this ethos? The shared premises just reviewed provide a starting point. First, the broadness of the definitions engenders a bothersome sense of circularity. The desired conclusion – that all action is planned, or, alternatively, improvised – is implicitly contained in the initial definitions of these very terms. Second, with regard to the quantitative-qualitative distinction, we should note that quantitative differences are nevertheless real differences. And if we look at points on the quantitative continua that are far away from each other, these differences appear to be important. Classical planning, for example, is brittle. It works only in artificially restricted environments. So-called continuous planning, in contrast, is resilient even in very uncertain, real world environments. Similarly, performing Beethoven’s Appassionata sonata is a far cry from playing free jazz with Sun Ra’s Arkestra. Indeed, vanishingly few musicians would even be capable of doing both. Moreover, we are not tempted to deny that red and blue are qualitatively distinct colours just because there is a continuum of shades between them on the continuous colour spectrum. In other words, quantitative differences can and do count as qualitative differences under some circumstances. So it is not clear that the quantitative-qualitative distinction cuts very much theoretical ice here.

These two attempts at assimilation, then, are not compelling. But are there principled reasons for rejecting any and all attempts at assimilation? We can take more inspiration from the colour example. In Western cultures, perceptual experience of red and blue is of two contrasting colours. This is a phenomenological difference based in part on biological equipment and in part on linguistic and cultural traditions.⁵ But it is a real, experienced, qualitative difference for those of us with standard issue colour perception machinery operating within these traditions. Similarly, I would argue, in these same traditions we have linguistic and cultural ways of distinguishing between planning and improvisation. While it is true that planning and improvisation share many features that vary only quantitatively between them, they are still phenomenologically distinct in the lived experience of acting in the world. Any attempt at assimilation of one to the other flies in the face of the phenomenology here.

Moreover, the phenomenological distinction between improvisation and planning is confirmed by the categorisations employed by scientists who study prospective behaviour – behaviour that concerns the future (Dickinson, 2011). The initial distinction is between habitual behaviour and goal-directed behaviour.⁶ Habitual behaviour is under the control of mechanisms like stimulus-response and reinforcement that ensure it will occur under the appropriate conditions regardless of the current motivational state of the animal. For example, your habit of fastening your seatbelt when you get into your car practically ensures that you will fasten it even when, having just remembered that you forgot to check your mail last night, you intend to get out of the car twenty seconds later at the end of your driveway to check it before proceeding to work. True goal-directed behaviour, by contrast, is under the control of current motivational states or goals. Dickinson (2011, p. 80) suggests it must also involve a sensitivity to the causal relations between the response choices available and the achievement of the goal. For example, when you get to the end of your driveway, rather than just braking as you usually do to ascertain whether it is safe to proceed into the street, you put your car in park, open the door, walk to the mailbox avoiding the puddles from last night’s rain, and so on. Alternatively, on seeing the puddles, you might brake, proceed into the street, and then back up so that you can reach the mailbox without getting out of the car. Whatever series of steps you take, though, the point is that you take them because you are currently motivated to achieve your goal and believe these steps will get you closer to it.

The distinction that is important for our purposes, however, is between goal-directed behaviour and future planning. Crucially, future planning involves behaviours that are not under the control of current motivational states but of future ones (Dickinson, 2011, pp. 85–86). If you call your favourite restaurant on Thursday to make a reservation for Saturday evening, it is not because you are hungry now. It is because you anticipate being hungry on Saturday. In addition, planning operates over longer time intervals than goal-directed behaviour – in the human case, not just days but weeks, months, years, or decades. So typically, not only do you have a goal you do not intend to achieve immediately, you also have in mind steps for achieving that goal which you do not intend to take immediately. Future planning, in other words, involves a significant delay between constructing a plan and executing it. The distinction between goal-directed behaviour and future planning is essentially the distinction between improvisation and planning/composition we have been discussing all along. Improvisation involves the pursuit of current goals, organising your activity as you go along in light of currently available resources.⁷ To plan, in contrast, is to pursue goals with a long view, organising your activity in advance in light of goals and resources you expect to have available in the future. We will return to studies of prospective behaviour in Section II of this chapter when we discuss how to describe the relationship between improvisation and planning more precisely. For now, the point is that attempts to assimilate them fly in the face of the distinction the behavioural sciences routinely make between goal-directed behaviour and future planning, just as they fly in the face of the phenomenology, and for similar reasons.

Finally, there are reasons of a different sort to resist assimilation stemming from evaluative concerns, rather than factual or interpretive ones. Val Plumwood argues for a distinction between dichotomy, which recognises difference, and dualism, which additionally subjects difference to differential valuation.

A dualism is more than a relation of dichotomy, difference, or non-identity, and more than a simple hierarchical relationship. In dualistic construction, as in hierarchy, the qualities (actual or supposed), the culture, the values and the areas of life associated with the dualised other are systematically and pervasively constructed and depicted as inferior.

(1993, p. 47)

Plumwood believes a system of interrelated dualisms – including culture/nature, male/female, and reason/emotion (p. 43) – undergird Western culture and explain the patterns of subordination and domination within it. The first element in each pair is the valued, superior one, and the second element the disvalued, inferior one. Moreover, there is typically an association of elements across the pairs – culture/male/reason are associated, as are nature/female/emotion, for instance. To this system of dualisms we should also add planning/improvisation.⁸ A clear example of how planning/improvisation fits Plumwood’s concept of dualism comes from the philosophy of music. Lee B. Brown (2011) notes that improvisation is actually endemic to musical practice and tradition the world over but has been increasingly suppressed and even denigrated in the Western classical tradition. In a particularly telling terminological choice, for instance, improvisation has been characterised as embodying an “aesthetics of imperfection” in contrast to the “aesthetics of perfection” embodied by composed music (Brown, 2011; Hamilton, 2000). Similarly, debate rages as to whether improvisers should be regarded as producing “works” of art at all, given the ephemeral nature of improvised pieces. The upshot is that the improvisational musical traditions of non-Western cultures – as well as non-white, non-elite cultures within the Western ambit – are routinely judged against the standards of the compositional tradition ascendant in elite, white, Western culture, and found wanting. Thus composition (p...