What do improvisers think is happening while they are improvising? How do some of them explain the cognitive processes that take place during a performance? In this chapter, the viewpoints of two prominent pianistimprovisercomposers, Frederic Rzewski and Robert Levin, will be analyzed and compared in order to present a continuum of the listening, forgetting, and conceptualizing that happens while improvising. Rzewski’s and Levin's improvisations in the contexts of music by Beethoven and Mozart will be discussed, in order to contextualize and substantiate their views towards improvisation, as well as to raise the question of how listening, forgetting, and conceptualizing could affect the way a performance sounds. Philosopher Donald Schön’s concept of reflectioninaction (Schön 1983, 68) as well as sociologist and amateur-pianist David Sudnow’s thoughts on embodiment (Sudnow 2001, 8384) will serve to reconcile the cognitive and physical aspects of performance by showing that they are in fact intertwined and inseparable.


Listening, forgetting, conceptualizing, playing


The question of what is happening in/with/through me while improvising has remained at the forefront of Playing Schumann Again for the First Time. While undertaking the work exposed in this project, I strongly believed that if I could understand what takes place (cognitively) during instances of musicmaking that exhibit many degrees of improvisation (as explained in the chapter On Scales of Improvisation), I could somehow use this knowledge to improve my improvisation skills. While previous chapters have shown how practice methods help musicians to develop a repertoire of materials that can emerge during improvised performances (see On My Improvisation Methodsand provided a detailed description of what drives the improvisatory process the feedback loop between mimesis and morphosis (see On Mimesis and Morphosis– this chapter addresses how the practitioner might be able to guide the improvisatory process during a performance. What are performers thinking and doing while improvising? In order to consider these questions, the perspectives of Rzewski and Levin here below will be taken into consideration thereby drawing a continuum of cognitive engagement with the concepts of listening, forgetting, and conceptualizing.

The most basic technique of improvisation is that of shortcircuiting this process of conservation: [...] forgetting  momentarily at least  everything that is not relevant to the objective of expressing an idea immediately in sound. […] Listening is the primary act. Performing is the secondary act. Every sound that one makes is a response to a sound one has heard. (Rzewski 2007, 52 and 102)


[Interviewer Frances-Marie Uitti]: Can you describe what you experience your brain to be actually doing [while improvising]?

Robert Levin: it is involved in a yin/yang between the conceptual and the muscular. If the fingers get too much ahead of the mind  or vice versa  there is a calamity. (Levin 2006, 509)


As can be seen from these quotes, Rzewski claims that the most basic technique of improvisation involves forgetting, while Levin references the duality between the conceptual and the muscular. I will combine their ideas in order to describe an approach to improvisation that involves listening to what has just been played, forgetting all the knowledge that one has accumulated through practice except that which is necessary for the current moment, and conceptualizing the constraints that are operative during the act of performing. These constraints take many forms: the duration of a performance, the idiomatic context in which the music can be understood, the particular constraints of a genre (the example of a four-voice fugue will be presented later in this chapter), the restrictions imposed by the presence of other players or by a particular piano, audience, or listening context, et cetera. Constraints are numerous and present in the case of all improvisation in performance. How can a performer listen, forget, conceptualize, and perform at the same time? To sonify the pursuit of an answer to this question, Rzewski's improvised cadenza during a live performance of Beethoven's Hammerklavier Sonata, Op. 106, will be presented alongside Levin's fantasy on themes by Mozart performed as part of his lecture “Improvising Mozart” in order to illustrate musically their respective opinions on this topic. In order to then bring some of the musical results of this research project into the discussion, the two versions of Lieder ohne Worte and the impact of the scenario surrounding their conception will be considered. Ultimately, Rzewski’s and Levin’s positions as to what is happening when a performer improvises will be reconciled by acknowledging that they are emphasizing different aspects of the truth. A discussion of Donald Schön’s concept of reflection-in-action and David Sudnow’s thoughts on embodiment will clarify that the cognitive and physical aspects of performance cannot be understood separately because they merge and influence each other in the “actionpresent” (Schön 1983, 62).1


Rzewski’s listening and forgetting while performing


In the article “Interpreting the Moment: On Improvisation and the Art of Forgetting,” Rzewski makes some general remarks about improvisation before presenting the idea that a basic condition for improvising is listening:


The improviser abandons self to the moment; abandons momentarily the world of necessity and causality for a world of possibility[; ...] things do not happen by necessity, nor by chance. They simply happen. Listening is the primary act. Performing is the secondary act. Every sound that one makes is a response to a sound one has heard. A solo improviser must respond to sounds made by self, as though made by another person. (Rzewski 2007, 102)


Rzewski’s thoughts about abandoning one’s self to the moment and leaving the world of "necessity and causality" for one of "possibility" induce questions regarding how these processes come about, and what the performer is doing in order to achieve this. He partially answers these questions in an ensuing article entitled “Little Bangs: Towards a Nihilist Theory of Improvisation” where he comes to the (perhaps ironic) conclusion that access to the world of possibility is gained through the art of forgetting:


The most basic technique of composition is that of transferring information from short-term memory to long-term: remembering an idea long enough so that one can write it down. [...] The most basic technique of improvisation is that of short-circuiting this process of conservation: forgetting - momentarily at least - everything that is not relevant to the objective of expressing an idea immediately in sound. (Rzewski 2007, 52)


Interpreting Rzewski's delineation between composition and improvisation as a distinction between composition (which, as shown in detail in the chapter On Scales of Improvisation is also improvisatory in nature) and improvisation in performance, a clear difference can be seen between writing down information and expressing an idea immediately in sound. I understand his notion of “forgetting” to be related to the process of letting the feedback loop between mimesis and morphosis guide improvisation in performance, a process described in detail in the chapter On Mimesis and MorphosisThe subchapter Eusebius Traum in the chapter On My Improvisation Methods contains three recorded performances, one of which could serve as a useful example of how forgetting relates to how the feedback loop between mimesis and morphosis guides performance. In two of the three performances, the constraints that I conceptualized before (as well as during) the performances were more or less fulfilled; but in the third instance, I forgot to use the designated melodic figurations and used other materials instead, which allowed for more room for the feedback loop between mimesis and morphosis to guide the performance.


Levin’s yin/yang between the conceptual and the muscular


In contrast with Rzewski’s idea that it is necessary to forget in order to be able to improvise, Levin describes how "the act of improvisation makes one aware at every point that there is always a multitude of possibilities from which to choose" (Levin 2006, 509) and emphasizes the practicing necessary in order to accumulate said possibilities. Rzewski’s claim about forgetting “everything that is not relevant to the objective of expressing an idea immediately in sound” (Rzewski 2007, 52) presupposes that a musician has access to such a multitude of possibilities that he or she has accumulated by regularly practicing various degrees of improvisation. (As discussed in more detail in the chapters On Rethinking the Current Performance Practice of Nineteenth-Century Music and On Scales of Improvisation, certain approaches to practicing classical music often do not engage with diverse improvisatory activities and rather restrict themselves to very few degrees of improvisation. Concrete suggestions as to how to integrate more degrees of improvisation into one’s practicing and therefore performing can be found in the chapter On My Improvisation Methods, specifically the subchapter Practicing.)

Levin also states that improvisation "narrows […] the distinction between spontaneous invention and what is on the page" (Levin 2006, 509). The way I understand this phrase is that he is describing how one comes to understand what other possibilities a composer might have had available at his or her disposal before choosing the one that is notated in the score, and likewise what possibilities a performer has at any given moment when a decision needs to be made during the course of an improvised performance. This is most likely an awareness that emerges from practicing improvisation regularly and not necessarily a realization that happens during any specific performed improvisation.

By speaking about the multitude of possibilities available while improvising Levin references the conceptual aspect of improvisation, the planning and conceiving that needs to take place for an improvisation to unfold over the course of its duration, a notion that he expands on when he explains the “yin/yang between the conceptual and the muscular” that he experiences his brain to be doing while improvising (Levin 2006, 509). “If the fingers get too much ahead of the mind—or vice versa—there is a calamity” (ibid.). He then describes the experience of improvising a Beethoven cadenza in concert and how the yin/yang relationship between the conceptual and the muscular affected his performance. Levin’s use of the term conceptual refers to a practitioner’s ability to channel the “multitude of possibilities” available to him or her in order to plan and conceive a successful improvisation, while the term muscular indicates the embodied knowledge gained through practice and used to realize a concept in performance.2 Considering that the brain is conceptualizing during the moment of improvising raises questions about how far ahead, or to what degree(s), the conceptualizing extends during the course of a performance. When exactly is the brain filtering through a multitude of possibilities and selecting the appropriate one for the moment? If too much conceptualizing takes place during the course of performance, the playing risks becoming decoupled from the cognitive processes taking place. By contrast, if embodied processes operate alone, there is no concept guiding the music-making. Finding a balance between these two aspects of improvisation seems to be what Levin is striving for when he describes the “yin/yang” relationship between them. Touching upon the necessity to relinquish control in order to find a proper yin/yang balance between cognitive activity and embodied processes, neurologist Aaron L. Berkowitz describes aspects of Levin’s performances in his book The Improvising Mind: Cognition and Creativity in the Musical Moment by using the term "letting go." Berkowitz explains that "once reaching the stage of 'letting go,' growing confidence in the ear, hands, and subconscious competence can allow the improviser to submit freely to the moment of performance. 'Letting go' means allowing the proceduralized/automatized subelements, processes, and structures of the knowledge base to guide the improviser from moment to moment" (Berkowitz 2010, 125). Berkowitz continues by saying that “Levin’s, extraordinary knowledge base, rich in both materials and their interconnections,” (ibid.) allows him to succeed while letting go in performance while simultaneously planning and conceiving that which needs to take place for an improvisation to unfold over the course of its duration. “Letting go,” in my opinion, describes the necessity to let go of certain cognitive activity, as well as certain embodied knowledge, in order to find the right balance between them in the moment of performance.




Listening, forgetting, and conceptualizing all happen during improvisation in performance. Returning to the discussion of the emphasis both Rzewski and Levin place on the physical aspect of improvising in performance, the question can be posed as to how listening, forgetting, and conceptualizing translate into embodied behavior, into the performing itself? In his book The Reflective Practitioner: How Professionals Think in Action, Donald Schön claims that


when someone reflects-in-action, he becomes a researcher in the practice context. He is not dependent on the categories of established theory and technique, but constructs a new theory of the unique case. […Because his experimenting is a kind of action, implementation is built into his inquiry. (Schön 1983, 68)


In the case of the improvising performer, the process of implementing an inquiry through practice negotiates the categories of established theory and technique theories and techniques resulting from the established “multitude of possibilities” with which the performer is already familiar and a new theory to fit the unique case, i.e. a specific musical moment during the course of an improvised performance. This process of negotiation takes place in the embodied act of performing. In other words, the “practitioner’s reflection-in-action […] is bounded by the ‘action-present,’ the zone of time in which action can still make a difference to the situation” (Schön 1983, 62), which in the case of the improvising performer is an extremely limited amount of time. Focusing more generally on how the embodied act of performing can be viewed by the practitioner, David Sudnow describes observing the muscular aspect of performance in his book Ways of the Hand, in which he explains the process of learning to play jazz at the piano while emphasizing the acquisition of embodied knowledge:


Looking down at my hands and finding a spate of jazz coming out, I’d find I was looking to the hands themselves, not their destinations, now seeing ways of travel over and above those particular notes being chosen. While before I’d looked past the hands’ ways to their destinations, […] the focal plane seemed closer, and I saw a configuring hand […] whose shaping was being watched, whose shaping and moving became gradually instructable. (Sudnow 2001, 8384)


The way Sudnow describes looking past the hands’ ways to their destinations sounds similar to Levin’s idea that too much conceptualizing and too little attention to the muscular can result in a calamity: because the hands are the tools that reach these musical destinations, they must remain involved in the process. By maintaining the position of “observer” a performer allows the movement of their hands to operate autonomously and without inhibitive cognitive interference. He then describes how he, while practicing, “expressly aimed for the sounds of these next particular notes, that their sounds seemed to creep up into my fingers, that the depression of the keys realized a specific sound I’d gone there to make” (ibid.). What is interesting in both Schön and Sudnow is that body and mind, the muscular and conceptualizing, are merging; this is what Rzewski is forgetting when he says that listening is the primary act and performing the secondary.


the first Lied ohne Worte performed on April 19th, 2016:

the second Lied ohne Worte performed on April 19th, 2016:

How does it sound?


As already mentioned, listening, forgetting, conceptualizing, and embodiment are operative in every improvised performance. The question will now be posed as to how giving priority (consciously or otherwise) to one or more of these three concepts can potentially influence the way a performance sounds? In order to musically illustrate an answer to this question, performances by Levin and Rzewski, as well as two of my own, will be discussed here in order to convey the idea that a tendency towards more listening and less conceptualizing (or vice versa) indeed influences certain audible parameters of a performance. Alongside my idea that all three concepts as well as the muscular aspect of playing (to use Levin's terminology), are intrinsic to improvising while performing, any particular concept can be given priority at any given moment. Although it is very difficult to show exactly how listening, forgetting, and conceptualizing affect a performance, precisely because they are all ever-present in every act of improvisation, the following discussion should initiate the idea that these concepts can render themselves audible.


How might the course of an improvisation be affected when a musician shows more interest in listening and forgetting than in conceptualizing? In order to arrive at an answer to this question, let me first turn to Rzewski’s performance of an improvised cadenza in the context of Beethoven's Hammerklavier Sonata live in Switzerland in 1991:

The cadenza is performed at the point in the score where Beethoven notates a fermata (towards the end of the first movement of the sonata), and it begins with more spacious-sounding material before proceeding to become moderately virtuosic at certain moments. I find this performance quite convincing on a number of levels: the harmonic and melodic language remain closely related to what Beethoven composed in the sonata, and the timing in the more spacious material conveys a sense of elegance and breadth, while the events in virtuoso passages come across as unexpected and spontaneous. From [2:46] to [2:56] Rzewski performs a diminuendo, immediately followed by a quick crescendo combined with an accelerando to return to the material from the opening bars of the sonata in Bb Major. I as a listener experience the music from [3:16] to [3:26] as exhibiting two or three different tempos. At circa [4:04] a very sudden shift from forte to piano can be heard. Two main questions arise when I listen to this performance and consider how I would like to be able to improvise in such a convincing manner: how, if at all, has Rzewski practiced in order to be able to play a cadenza like this? And what is he thinking about during the course of the performance? An assumption that Rzewski does practice (based on having witnessed him do so upon the many times I visited him at his home) coupled with Derek Bailey’s statement that, "with solo improvisation [...] there are definitely possibilities for practise" (Bailey 1992, 110), at least partially addresses the first question. With regard to the second question I need to return to Rzewski's claims that the most basic technique of improvisation is "forgetting everything that is not relevant to the objective of expressing an idea immediately in sound," and connect this to his other statement, that listening is the primary act and performing the secondary. However, a new question quickly arises: how might this perspective towards improvisation affect the way his improvising sounds?

Cognitive scientists Annerose Engel and Peter E. Keller assume in an article entitled “The Perception of Musical Spontaneity in Improvised and Imitated Jazz Performances” that “the ability to judge whether a performance is improvised or rehearsed depends on the listener’s sensitivity to fluctuations in parameters such as event timing and intensity" (Engel and Keller 2011, sixth paragraph of the Introduction). They then describe an experiment they undertook where experienced jazz musicians played “spontaneous improvisations” and “rehearsed performances” (Engel and Keller 2011, Stimuli) and listeners were asked to determine which was which. The results show that “spontaneously improvised piano melodies are characterized by greater variability in timing and intensity than rehearsed imitations of the same melodies, and highly experienced, empathic listeners can detect these differences more accurately than expected by chance” (Engel and Keller 2011, first paragraph of the Conclusion). To clarify, Engel and Keller pinpoint the term intensity as “loudness” (Engel and Keller, first paragraph of the Introduction). In order to use the findings of their study to partially answer the question of how Rzewski’s perspective towards improvisation might affect the way his improvising sounds, I will soon debate their juxtaposition of improvised and rehearsed modes, rather understanding performances on a spectrum where they all exhibit rehearsed elements and degrees of improvisation (as shown in detail in the chapter On Scales of Improvisation). Then I will use three further musical examples in order to show that the ratio of listening, forgetting, and conceptualizing is in fact (one of) the determining factors as to how timing and intensity (loudness) are realized in performance. I will turn to one of these examples now before returning to the debate regarding juxtaposed and improvised melodies.

Just as one might question how listening and forgetting may affect the way an improvisation sounds, and in order to account for Levin's inclination to discuss the conceptual aspect of improvising, the following question could also be posed: how might conceptualizing affect the way an improvisation sounds?

At the end of his lecture “Improvising Mozart,” hosted by the Centre for Research in the Arts, Social Sciences, and Humanities at the University of Cambridge (UK), Levin concludes with this improvisation in the style of Mozart:

Levin begins conceptualizing the improvisation before the playing by asking the audience for thematic suggestions from the Viennese classical repertoire (Levin Improvising Mozart 2012, [1:01:131:01:30]), in order to take materials (melodies and harmonies) from their suggestions and use them during the performance. His strategic request that the suggestions are by Mozart, Haydn, or early Beethoven helps the suggested material align to certain stylistic parameters. Members of the audience provide him with the following suggestions, all by Mozart: the opening bars of String Quartet No. 19 K. 465, the Rosenarie from Le Nozze di Figaro K. 492, and the principal theme from the first movement of Piano Sonata in F Major K. 332. Levin comments that it is quite convenient that K. 332 and the Rosenarie are both in F Major, presumably so that he does not have to concern himself with transposition, and proposes reordering the audience’s suggestions so that the key scheme is more sensible: an F Major / C Major / F Major structure commonplace to the repertoire in question.3 My analysis of the performance is that it contains a prelude (until [1:04:19]), which, along with the transitions between the three thematic segments reflecting Mozart's scores, exemplifies the largest degrees of improvisation in the performance. After the prelude and between the transitions the ensuing music could be considered a presentation of Mozart's materials from the respective scores, memorized by Levin when the performance does remain compliant to certain excerpts of a specific musical work, and altered when he decides to depart from the source material either because he is exploring another possibility than what ended up in Mozart's score or because he is creating developmental passages and transitions to the next material suggested by the audience. Although the rhythmic and dynamic aspects of the piano playing exhibited in the performance are relatively predictable, Levin’s ability to perform music related to Mozart scores by memory based on suggestions from the audience certainly demands a high degree of conceptualizing. Levin does perform a diminuendo and ritardando from [0:34] to [0:42], but when he begins to perform one of the audience’s musical suggestions (at circa [0:42]), the rhythmic and dynamic elements of his performance remain more stable. To what degree is Levin conceptualizing during this performance, and how much is he also simultaneously "listening and forgetting?" And, related to the question I proposed before of how conceptualizing affects the way a performance sounds, how does his conceptualizing as well as his listening and forgetting affect the way his performance sounds? Two more musical examples presented here below will assist in answering these questions.

Levin’s free fantasy exhibits different approaches towards the musical parameters of timing and dynamics than Rzewski’s cadenza. These differences are a result of many factors, including their respective sonic signatures (see the chapter On Sonic Signature), and as I will now argue – the respective ratios of listening, forgetting, and conceptualizing they exhibit during performance. Engel and Keller’s conclusion that “improvised piano melodies are characterized by greater variability in timing and intensity than rehearsed imitations of the same melodies,” contextualized by the knowledge that improvisation and rehearsing are not opposite ends of a spectrum, will help with this argument. That listeners were accurately able to judge whether the musicians in the study were in fact improvising or performing rehearsed melodies (Engel and Keller 2011, Behavioral Data) reflects a clear difference in these two activities. A more detailed understanding of what improvising and performing rehearsed melodies involve is now necessary in order for me to reject Engel and Keller’s juxtaposition of “improvised” and “rehearsed” in favor of activities that all involve listening, forgetting, and conceptualizing, whereby the ratio of the three differs. While improvising melodies, the performers were relying on access to a multitude of possibilities, some (or even most, or perhaps all) of which had already been rehearsed (practiced) before in order to become integrated into their embodied knowledge. When “each of the six pianists returned to the laboratory to imitate the selected excerpts from his or her improvisations (self-imitation) and those produced by two of the other pianists (other imitation)” (Engel and Keller 2011, fourth paragraph of Stimulus Generation), their reliance on a multitude of possibilities while performing a rehearsed melody became overshadowed by another demand, namely the necessity to imitate a past performance. Although the performers, while fulfilling the task of imitating a previously occurring performance, had access to their same individual multitudes of possibilities as when they improvised the melodies, a new layer of conceptualizing was necessary in order to attempt to re-create a past performance. Based on the findings of Engel and Keller’s study, I theorize that this added demand to conceptualize further constraints (not just a melody of a certain length in the context of a specific jazz idiom, but also a re-creation of such a melody that had been previously performed) limited the amount of forgetting and listening the performers were able to engage with, and this in turn affected the timing and loudness of their performances.4

A similar situation can be found in Playing Schumann Again for the First Time. At the beginning of a working session with Rudolf Lutz on April 19th, 2016, I was asked to play a Lied ohne Worte without very much further instruction; and, about two hours later, after the private session and during a more concertlike scenario where more people were present, I was asked to "recreate" the first Lied ohne Worte. In the meantime, I did not have the opportunity to listen to the audio recording.

Listening back to the recordings of both performances, the element that varies most widely between these performances relates to timing. Although the timing in both performances is relatively flexible, the first performance is more expressive in that it balances more risktaking in terms of timing events with a convincing flow throughout the performance. In the first performance, after the melody note at around [0:29], the accompanying material is performed later than expected; and a ritardando can be heard from around [0:46–0:48]. In the second, a ritardando is performed from [0:31] to [0:34]; timing events become unpredictable around [0:53–1:02] because of the performance of certain wrong notes; and a ritardando and diminuendo can be heard from [1:30] to [1:32] before the performance of the coda. Many factors contribute to these differences, but an explicit factor in this case is the challenge to recreate the first performance as exactly as possible. I claim that such a scenario demands a higher degree of conceptualizing and therefore can potentially restrict the performer’s ability to let the performance be guided more by listening and forgetting. Whereas conceptualizing, as well as listening and forgetting, assisted in the process of creating the music in both of these moments, the requirements of the second scenario – to refer back to a previous musical event – involve both the conceptualizing, listening, and forgetting necessary to create a musical moment, as well as the necessity to remember a previous event in order to attempt to recreate the past. While performing these two songs without words, I felt more able to let the feedback loop between mimesis and morphosis guide the first performance, due to the fewer demands to conceptualize during the course of the performance.


One more answer to the question of how listening, forgetting, and conceptualizing affect the way a performance sounds relates to the restrictions a certain genre places on the performance. For example, there is clearly much less room for forgetting and more necessity for conceptualizing while improvising a fourvoice fugue according to historical contrapuntal standards than while improvising a free fantasy. Forgetting does also certainly play a role, as it does in all improvised performance, but the technical demands of improvising counterpoint demand a high level of conceptualizing to take place during performance, thereby affecting the ratio between listening, forgetting, and conceptualizing.

These two responses as to how listening, forgetting, and conceptualizing affect the way a performance sounds offer partial answers to the questions of how Rzewski and Levin’s respective listening, forgetting, and conceptualizing affect the way their improvised performances sound. The assumption could be made that, while Rzewski prefers to mention listening and forgetting as the main driving concepts for improvisation in performance, he tends to prefer listening and forgetting over conceptualizing. And in turn, because Levin prefers to discuss conceptualizing, he tends to conceptualize more than he listens and forgets while performing. This assumption may however be unnecessary in lieu of Engel and Keller’s findings, which correlate diversity in timing and loudness to improvised performances. I have argued in this subchapter that a specific ratio of listening, forgetting, and conceptualizing (where listening and forgetting is more prevalent than conceptualizing) can affect the timing and loudness parameters that are audible in performance, and better allows the feedback loop between mimesis and morphosis to guide a performed improvisation.


To conclude, I would like to suggest that the readerlistener consider how listening, forgetting, and conceptualizing always and necessarily affect the course of an improvised performance as well as how it sounds, in order to reengage with the music presented in this project. If the readerlistener is a practicing musician, I encourage him or her to consider these three concepts while improvising, in order to see how a different ratio between them might create a different engagement with the feedback loop between mimesis and morphosis that always also guides improvisation in performance.