1. SPIRALLING OUT

Today, the question of music is entangled with that of machine listeningFootnote ¹ within a more-than-human ecology of sound, to the extent that they are both situated in the convergence of proliferating apparatuses of dataficationFootnote ² and post-anthropocentrism. Musical experience is almost unavoidably processed by the mechanisms of surveillance capitalism, capturing the behaviour of human and non-human lives as data (including music-making/consuming) in order to produce economic surplus through feedback loops of behavioural prediction and modification (Zuboff Reference Zuboff2019). Under these conditions, perception and datafication are co-constituted, ‘troubl[ing] rigid boundaries between machine and human listening’ (Amsellem Reference Amsellem2024: 28), which in turn troubles the notion of music defined as exclusively human. At the same time, listening machines seem to be indifferent to the human as exceptional, since they are equally applied to datafying animal behaviour through sound (Lupton Reference Lupton2023: 82). In this way, they are part of a post-anthropocentric gesture which ‘displaces the notion of species hierarchy and of a single, common standard for “Man” as the measure of all things’ (Braidotti Reference Braidotti2013: 67). As posthumanist philosopher Rosi Braidotti points out, today ‘no animal is more equal than any other, because they are all equally inscribed in a market economy of planetary exchanges that commodifies them to a comparable degree and therefore makes them equally disposable. All other distinctions are blurred’ (ibid.: 71). In this sense, Jeremy Bentham might have been right to predict a time ‘when humanity will extend its mantle over everything that breathes’ (Bentham cited in Cimatti Reference Cimatti2020: 187), except that he missed the part where humanity itself (defined in terms of a double negation: non-non-human) is dismantled. The world of machines, thus, contains the possibility of dissolving the boundaries between human listening and animal hearing, allowing us to think of hybrid, more-than-human ways of being in sound.

How could we approach the question of musical listening beyond the binary of nature/culture, attending to ‘animals [that] qualify as cyborgs, that is to say as creatures of mixity or vectors of posthuman relationality’ (Braidotti Reference Braidotti2013: 73)? Let us think of a real creature, half animal half machine, that motivated the research presented in the following pages: a mollusc bio-sensor (Distributed Sensing Systems Group 2020). To produce the ‘best oyster’ for human consumption, farmers in Tasmania have designed cyborg oysters with implants of digital sensors, aiming at optimising their production by correlating the creatures’ biometric data with the environmental data (Lupton Reference Lupton2023: 87–8). While this method is remarkably close to the science fiction (SF) of the past – particularly the ‘super-mollusc’ produced by the fusion of computation and biology, imagined by Vilém Flusser in 1988 (Flusser Reference Flusser2011: 144–5) – it is equally relatable to today’s human society, as it appears that ‘a happy oyster is a big-data oyster’ (Rutkin Reference Rutkin2014). The figure of this humanised machinic mollusc is conducive to what Donna Haraway refers to as SF, a method of (un)making ‘multispecies’ worlds through the entanglement of speculative fabulation and science fact (Haraway Reference Haraway2016: 2–3). In her recent work, the scope of SF consists of ‘science fiction, speculative fabulation, string figures, speculative feminism, science fact, so far’ (ibid.: 2–3). The present research attempts to link Kodwo Eshun’s methodology of Sonic Fiction with SF, to develop a critical, post-anthropocentric approach to machine listening.

Sonic fiction is a theoretical framework developed by Kodwo Eshun’s investigation of ‘an unofficial mythology’ of ‘the machine as a lifeform’, situated in the twenty-first century ‘discontinuum of AfroDiasporic Futurism’, ‘oppos[ing] common sense with the force of the fictional and the power of falsity’ (Eshun Reference Eshun1998: A[179], A[189], 00[-003]). It is a potent framework for approaching the post-anthropocentric conceptechnics Footnote ³ of machine listening, through the method of sonic thinking. Traced by Holger Schulze, this way of thinking with and by means of sound works through the tripartite structure of mixillogic, mythscience and mutantexture (Schulze Reference Schulze2020: 19–22). The first term signifies the kind of logic characterised by ‘joyful transgression’ that is made ‘out of mixtures’ and ‘applies steps in thinking that would not be regarded adequate in scholarly logic’ (ibid.: 21). It is a logic that, contrary to common sense, is adequately equipped for the transversal terrain of the posthuman. Coined by Sun Ra, the second term is described as ‘a new and apparently mythologically structured or grounded kind of scientific knowledge’, with the capacity of going against reductive traditional epistemologies (ibid.: 19). In relation to the problem of control posed by contemporary networks of surveillance and datafication, Simon O’Sullivan develops mythscience as a practice of fictioning, where art is mobilised as a form of resistance against the contemporary homogenisation of subjectivity (O’Sullivan Reference O’Sullivan2014, Reference O’Sullivan2016). The third term describes the effect of the other two, that is, a material sonic trace of ‘a diffracted logic’, productive of more expansive auralities and plural possible worlds (Schulze Reference Schulze2020: 21, 35–40). Thus, sonic thinking allows us to listen in the present mytho-techno-scientific reality, seeking for alternate logics as points of access to sonic (im)possibilities.

What follows is a critique of the homogenising conceptechnics of machine listening through a process of fictioning, engaging with the sonic thinking that unfolds around the figure of the mollusc bio-sensor as a locus of a malacological mythscience.Footnote ⁴ This involves the development of a molluscan mixillogic applied in the composition of a more-than-music assemblage, productive of molluscan mutantextures.Footnote ⁵

2. ANTHROPO-LISTENING MACHINES

The idea of a listening machine is not at all new to music.Footnote ⁶ Most tech-friendly artists (or even consumers) are routinely in contact with machines that are listening in one way or another, although there is little consensus among researchers and practitioners about the definition of machinic listening (Sterne Reference Sterne2022). A pragmatic definition of machine listening would be a digital system that involves the extraction of meaningful data from audio streams and its insertion into machine learning as input data and/or training sets. However, even though more subtle definitions diverge widely, there seems to be a consensus when we look at the presuppositions within discourses that justify various design and development trajectories.

One of the earliest uses of the term ‘machine listening’ appears in the work of the musician and engineer Robert Rowe (Stern et al. Reference Stern, Dockray and Parker2021: 25:54). In his seminal study on computer music, Interactive Music Systems, he states that we should ‘follow the example of human cognition in building a machine listener [because] humans provide the best, indeed the only, model of successful music listeners [and thus, the music produced] will resonate with the human cognitive processes for which it is ultimately intended’ (Rowe Reference Rowe1993: 119; emphasis added). This claim to human exceptionalism in relation to music recur like a refrain in various definitions of machine listening until today. For instance, Michael Casey writes in the Oxford Handbook of Computer Music that ‘an important area of research in the last decade or so has been machine listening; the goal is to make computer systems that are capable of humanlike hearing and decision making’ (Casey Reference Casey and Dean2011: 450). Andrew Hugill defines them as ‘models of the [human] auditory systems’ adding that ‘there remains a major challenge in achieving a machine listener that “understands” the causal way in which human beings hear in real time’ (Hugill Reference Hugill2019: 26). Finally, Thor Magnusson claims that:

Instead of a search for novelty, innovation and the affirmation of difference, the history of machine listening reveals a tendency to reproduce a presupposed ideal model of human listening, even though it is very hard to determine this model in the first place. The Aural Diversity project has effectively deconstructed this normative listening experience, which has been the ground of most of the acoustic design in Western societies, despite being at best a reductive and exclusive fiction (Drever and Hugill Reference Drever and Hugill2023), or an ‘analytic fact of Nobody’ (Deleuze and Guattari Reference Deleuze and Guattari2013: 123). Furthermore, Jonathan Sterne points out that a problematic and anthropocentric notion of listening is a common ground for most of the debates around machine listening, adding that currently ‘machine listening is intimately connected to corporate attempts to enclose more and more domains of human interaction, and to state surveillance and authoritarian projects in many parts of the world [and thus] any theory of the listening in machine listening needs to also be a theory of power’ (Sterne Reference Sterne2022: 3). This resonates with the collective research project Machine Listening, a Curriculum, which defines machine listening as ‘much more than just a new scientific discipline or vein of technical innovation … It is also an emergent field of knowledge-power, of data extraction and colonialism, of capital accumulation, automation and control’, developing into a political field that ‘demands critical and artistic attention’ (Dockray et al. Reference Dockray, Parker and Stern2024). The present research follows precisely these lines into the discourse and practice of machine listening, by considering the ‘listening machine’ simultaneously as another machine: the biopolitical apparatus described by Giorgio Agamben as the anthropological machine.Footnote ⁷

In The Open: Man and Animal, Agamben develops his theory of anthropogenesis, the production of the Western notion of the (hu)man, as the result of ‘a caesura and articulation between human and animal’, where ‘Man suspends his animality and, in this way, opens a “free and empty” zone in which life is captured and a-bandoned {ab-bandonata} in a zone of exception’ (Agamben Reference Agamben2004: 79). Agamben describes this process of distinction between the human and the non-human as the anthropological machine and calls for critical attention: ‘We must learn instead to think of man as what results from the incongruity of [the human and animal], and investigate not the metaphysical mystery of conjunction, but rather the practical and political mystery of separation. What is man, if he is always the place – and, at the same time, the result – of a ceaseless division and caesurae?’ (ibid.: 16). In short, his analysis defines the anthropological machine as an evolving process of exclusion. Each iteration or assemblage of the anthropological machine produces the propriety of the human by proposing a critical characteristic for the judgement of human nature. Agamben primarily focuses on language as one of the most consistent characteristics that have been sustaining this distinction. He investigates the figures of the Pithecanthropus Alalus (literally translated as the ‘speechless ape-man’), whose existence is demarcated right outside the evolutionary threshold of the ‘speaking man’ (ibid.: 34), and the Homo Ferus, a name given to the enfants sauvages, human-born children grown up among animals in the eighteenth century. Their muteness (due to their linguistic lack) and facelessness (as they may not have walked on two legs, thus concealing their head), excluded them from the species Homo Sapiens (ibid.: 30). This time, in reverse, we see that the anthropological machine turns into a listening machine – the anthropo-listening machine – as it has an ear for the silence, muteness and illegibility of bodies, only to draw an anthropogenetic line around the human through the exclusion of everything else.

In our times, narratives and practices of AI, including machine listening, play an important role in sustaining such mechanisms of exclusion. There are numerous cases where the ‘black boxes’ of neural networks have (re)enacted unacceptable biases (Hern Reference Hern2018) while critical attention has been aimed at the violence and perils these biases might entail (McQuillan Reference McQuillan2022). Following the case of language, it seems that AI easily slips into an anthropological exclusion based on Intelligence – or the kind of behaviour it is designed to simulate in each iteration – as the defining characteristic of the human. Contrary to these anthropocentric narratives of artificial intelligence, the artist, writer and technologist James Bridle sees the potential of a more-than-human and critically ecological approach to these technologies, for attending to the plurality and diversity of the planetary intelligence(s) defying anthropocentric boundaries. In Ways of Being, Bridle shows how ‘intelligence is ecological – that is, entangled, relational, and of the world’, which in turn implies that ‘artificial intelligence provides a very real way for us to come to terms with all the other intelligences which populate and manifest through the planet’ (Bridle Reference Bridle2023: 57). Could we extend this view to AI ways of listening?

Despite Rowe’s assumption that ‘musical listening’ is exclusively human, there is a plethora of sonic artefacts and mutantextures that testify to the workings of transversal musical sensibilities and mixillogics that put to question the human exceptionalism associated with musical listening. The work of Florian Hecker, among others, exhibits machine listening’s capacity for de-centring the rigid model of the human – upon which the anthropological machine relies – from the potentialities of musical listening. Iteratively moving between analysis and synthesis, Hecker interrogates state-of-the-art machine listening technologies, attempting ‘to open a dialogue with these spectral operators … in order to render audible their intelligence signature, the signal trace of their nonhuman brain-ear’ (Hecker and Kronic Reference Hecker, Kronic, Bonnet and Sanson2021: 57–8). Through this process, described as ‘synthetic listening’, Hecker’s work attempts to reach the aesthetic dimension of sonic phenomena that ‘no human listener would ever happen upon’ (ibid.: 63). The idea of the brain-ear, developed by the philosopher and Hecker’s collaborator Maya B. Kronic, is premised on viewing the human sensorium as a ‘patchwork’ (Kronic Reference Kronic2015), akin to a machinic assemblage, or ‘agencement’, where the human is always entangled and potentially inseparable with non-human entities and events (Miyazaki Reference Miyazaki2012). Diving into Hecker’s sonic fiction, the figure of the chimera Footnote ⁸ appears in a variety of pieces and forms that ‘the listener must reconstruct as a unified yet impossible synthetic creature’ (Kronic Reference Kronic2015). The impossibility of Hecker’s sonic chimeras can range from the mostly unintelligible Articulação (2014), to the machinic ‘intelligence structure’ that produced Synopsis Seriation, which ‘does away with [the former] staging of computer-generated speech [in favour of dramatising] synthetic sound in all its unnameable intensities’ (Hecker Reference Hecker2021b). The mutantexture that emerges from Hecker’s chimeras has entered the AI terminology under the label of the ‘templetexture’: a notion that ‘calls for machine listening systems that do away with any recourse to “coarse” semantic labelling’, by amplifying the humanly imperceptible incongruences between semantic machine listening, auditory neurophysiology, and sonic processes of texture synthesis (Hecker and Kronic Reference Hecker, Kronic, Manouach and Engelhardt2022).

Apart from the audible content, the chimera can be adequately used as a metaphor for the listening subject encountering Hecker’s sound-worlds – some to which ‘only machines can listen’ (Hecker Reference Hecker2021a). Given the understanding of the human sensorium as a patchwork that remains integrated enough to produce a coherent, intersubjective ‘image of the world, and of the subject that inhabits it’, Kronic proposes that ‘it is also possible to engineer circumstances in which this integration fails’ (Kronic Reference Kronic2015). Counter to the aesthetic ideal of disinterested pleasure, Hecker’s work (re)synthesises the listener through a ‘non-integrable or disintegrated experience’ (ibid.) – akin to what Deleuze would call the ‘transcendental exercise of the faculties’ – recognising its potential to break down the Kantian myth of the unified subject through aesthetic experience (Cox Reference Cox2018: 212–13). As Kronic puts it, Hecker’s work replaces the auraltypical subjects of psychoacoustics with ‘a laboratory humanity … populated by disintegrated and hallucinating subjects’ (Kronic Reference Kronic2015). However, this approach runs the danger of producing a mode of listening that is too abstract (or not abstract enough), in the sense that it fails to condition the continuity required for more-than-human entanglements. Instead of a lab-logic approach to listening where the first step is to be ‘isolated’ (ibid.), the foundation for an ecological machine listening is radical relationality.

George Lewis offers an exemplary approach to machine listening, situated as a way of resisting ‘dominant discourses and research directions in the field [of strong AI, as well as computer music, that] continue to assume a Euroamerican model of human subjectivity’ (Lewis Reference Lewis and Dean2011: 461). His research shows how the ‘human’ model of machine listening is far from a neutral and natural term; but rather a term constructed through racialised power relations, relying on an exnominated whiteness (Lewis Reference Lewis2002: 99–103), or a ‘white aurality’ defined through the normalising exclusion of other possible sensibilities (Thompson Reference Thompson2017: 273–5; Schulze Reference Schulze2020: 61–5). Instead of relying on prefabricated notions of listening and musicality, his revolutionary Voyager (Lewis Reference Lewis2000b; Lewis and Mitchell Reference Lewis and Mitchell2019), realised in 1988, is a sonic experiment of how radically different creative intelligence can be, and how we could make sense with it (Lewis Reference Lewis2000a: 38). The answer is given through the sonic persona of the ‘creative machine’ and the ‘Afrological’ notion of improvisation. While resisting the privilege of its ‘Eurological’ counterpart, composition, this approach grants the machine a mode of listening capable of ‘dialogic imagination’, thereby offering great potential for disentangling the anthropo-listening machine (ibid.). Defined as the ‘bidirectional transfer of intentionality through sound’, emotional transduction offers a way to conceptualise the vibrant relationalities possible in a sonic ‘technology-mediated animism’, and differential musical processes of interactivity where the ‘[human] identities are continually conditioned and reinscribed’ (ibid.: 36–8). In short, by emphasising the situated agency of machinic listening assemblages by inscribing ‘values such as relative autonomy, apparent subjectivity, and musical uniqueness rather than repeatability’, Lewis (Reference Lewis, Dean and McLean2018: 127) shows us how to allow creative machines to teach us how to listen otherwise.

The emergence of different ‘synthetic’ ways of listening is generative of a diverse Synthetic Acoustic Ecology (SAC) that might promote ways of being in sound on par with non-human others. Eldridge and Kiefer view such an ecology as a potent lingua franca for bridging empirical/theoretical, cultural/biological and artistic/scientific sonic epistemologies (Eldridge and Kiefer Reference Eldridge and Kiefer2018). By promoting ‘sonic situatedness’ and ‘sonic embodiment’, they view the practice of building audio-based sonic worlds of interactive agents as a way of deepening our responsiveness to the soundscape’s emergence. Their proposition for such a framework is primarily intended for monitoring and prediction applications, aiming for an ecological optimisation according to the premise that the ‘soundscape structure is a proxy for ecological integrity’, while warning us ‘to avoid projecting added reality onto simulated models’ (ibid.: 6). However, turning SAC into a predictive system of optimisation seems to contradict the former ambitions. In the words of the feminist technoscience researcher Thao Phan, ‘while ostensibly oriented towards the future, in practice, predictive models operationalise the past. At their worst, these systems don’t only perpetuate inequality but foreclose alternate possibilities for futures that do not conform to previous patterns and expectations’ (Phan Reference Phan2022: 54:36). In the case of SAC, this should remind us of immanent bias, to the extent that it exclusively assigns ecological relevance to certain actors and actions and not others. For instance, how would it include creatures that are silently affected by sound, if the sonic data only describe vocalisations? We can find an abundance of examples in the vastness of the ocean, where, apart from the usual suspects of like-large-brained cetaceans, there is a plethora of creaturely sonic entanglements that are deemed mute by anthropocentric onto/epistemologies (Vermeij Reference Vermeij2010). This does not mean that they should be excluded from the scope of a possible synthetic acoustic ecology, but rather we should learn from their strangeness and amplify our efforts towards attending our ears to the inaudible. Otherwise, the anthropological machine will keep feeding on the muteness of the sonically Other. Instead of exnominating the legitimising process of pure data, creative practice might offer more through sonic fiction precisely by actively and critically engaging with that added reality, since, according to Phan, ‘our legitimacy is in our excess: our failure to align with the homogeneous technical standards’ (Phan Reference Phan2022: 01:06:37). Thus, a truly ecological listening might emerge through diffractive mixillogics rather than synthetic logics.

3. (DIS)MANTLED SOFTNESS

Having established the genealogical entanglement between machine listening and the anthropological machine, what follows is a ‘strategic alienation’ (Schulze Reference Schulze2020), or a process of unbecoming through sound, seeking ways of dismantling the anthropo-listening machine. Felice Cimatti, a contemporary philosopher who explores the possibilities of such an endeavour, writes that the first step is ‘to be exposed to the animal’s gaze’, and the second is ‘to accept the possibility of a transformation, contained in the animal’s gaze’ (Cimatti Reference Cimatti2020: 96–103). In this case, the gaze is far from limited to an anthropomorphic vision; it ‘has nothing to do with the eyes, but with the fact that something – something nonhuman or even inhuman – is staring at us, whether or not it has eyes’ (ibid.: 96–100). In fact, the more surprising the gaze, the more effective the rerouting of the anthropological machine into becoming-animal.Footnote ⁹

In Matthew Calarco’s work on critical animal studies, becoming-animal is defined as ‘a refusal to enact the ideals and subjectivity that the dominant culture associates with being a full human subject and to enter into a relation with the various minor, or non-dominant modes of existence that are commonly viewed as being “other” than human’, while ‘the animal often serves as the chief limit against which human propriety is instituted’ (Calarco Reference Calarco2015: 57). The conditions capable of setting forth such a process are often described as ‘zones of indiscernibility’ (Deleuze Reference Deleuze2017: 15–19), ‘zones of intensity or proximity’ (Deleuze and Guattari Reference Deleuze and Guattari2013: 319), ‘zones of exception’ (Agamben Reference Agamben2004: 79), or ‘zones of indistinction’ (Calarco Reference Calarco2015: 57). The latter is fundamental in Calarco’s theory of ‘indistinction’ (ibid.: 48–69), where he explains how such zones produce a space for dismantling the anthropological machine, while challenging anthropocentric structures of power and amplifying the ‘more-’ that has been excluded from the notion of the human as described earlier (ibid.: 62–3). The capacity of listening, machinic or not, for opening such zones of indistinction is illustrated in Espen Sommer Eide’s ‘Exercises in Non-human Listening’ (Eide Reference Eide, Bonnet and Sanson2019), where he speculates on music as becoming-mosquito. Fascinated by the way of hearing practised by the mosquitos he encounters in northern Norway, he envisions a musical instrument that can simulate the way mosquitos listen to each other.Footnote ¹⁰ Furthermore, he draws inspiration from the animism of the Yukaghir people, whose hunters ‘lose themselves and become the animal completely’, entering ‘an in-between state, between the animal and the human’ by embodying an ‘imperfect copy’ of that animal (ibid.: 30). More accurately, Deleuze and Guattari stress how this imperfect copy should not follow a logic of imitation or representation (Deleuze and Guattari Reference Deleuze and Guattari2013: 319), but one enters a zone of indiscernibility by allowing their body to ‘enter into composition with something else’, and further, that ‘this something else can be quite varied, and be more or less directly related to the animal in question’, referring to the examples of the ‘animal’s natural food’, its ‘relations with other animals’, or an ‘an apparatus or prosthesis to which a person subjects the animal … in question’ (ibid.). In the case of becoming-mosquito, that something else is the ‘distorted ghost tone’ that enables the mating mosquitos to relate with each other into a ‘tuning duet’ (Eide Reference Eide, Bonnet and Sanson2019: 29). Such a process would reroute the anthropo-listening machine through which ‘we pick out tonal signals and discard the noise’ by composing as ‘the mosquito tunes into the noise and discards the signal’ (ibid.: 30). So, the third step for dismantling the anthropo-listening machine appears to be a matter of selection, of picking up and following the thread of difference (or ‘something else’) in relation to an animal.

In Gerard Raunig’s work, softness is framed as a thread leading out of an exclusive, masculine subjectivity characterised by domination, appropriation and possession (Raunig Reference Raunig2024: 86–90). He develops softness as a way of relating through affections ‘with all kinds of critters, with plants and animals, with technical apparatuses and thing-worlds’, capable of producing subjectivities that go ‘beyond anthropocentric perspectives’ forming what he terms the ‘human-machinic soft-ware’ (ibid.: 90, 79). As he shows in a letter to Suely Rolnik, this notion of softness was already articulated by Felix Guattari in the question of becoming-animal, when the latter envisions a ‘multiplicity of mutant becomings … as so many ways of inventing, of “machining” new sensibilities, new intelligences of existence, a new softness’ (Guattari Reference Guattari2011, cited with a slightly altered translation in Raunig Reference Raunig2022: 79) – or, a new softness between humanity and animality. Thus, the question of becoming-animal is already entangled with the properly (malaco)logical question of becoming-soft.Footnote ¹¹ Could we then dismantle the anthropo-listening machine by composing with the ‘gaze’ of animals defined by their softness (molluscs)?

4. MEMORIES OF QFWFQ

In the work of Deleuze and Guattari, the question of becoming-animal is intimately linked with literary fiction, which offers methodologies for building routes of escape from exclusively ‘human’ ways of living (Burrows and O’Sullivan Reference Burrows and O’Sullivan2019: 25). In fact, an encounter with a fictional mollusc was the pretext for applying these questions in practical experimentation. During Italo Calvino’s centenary (2023), I was asked by pianist Carolina Cury to compose a piece for piano and machine listening/synthesis,Footnote ¹² based on Italo Calvino’s Cosmicomics (Reference Calvino2010). This collection of short stories, originally published in Italian in 1965 and in English in 1968, weaves a SFFootnote ¹³ cosmology propelled by the constant metamorphoses of the main character, Qfwfq. Each story starts with the presentation of a scientific factoid which is extrapolated into (im)possible worlds of more-than-human becomings. The music composition is titled The Spiral, after Calvino’s homonymous story, which begins with the following passage:

Immersing myself in Qfwfq’s molluscan experience, I was drawn to the paradoxical process where the molluscs produce sensations that are imperceptible to them, but nevertheless enable them to enter into meaningful and aesthetic relations with totally other entities. The metaphor employs the sensation of the image appearing on the mollusc shell and the sense of vision, which is not granted to molluscs by the scientific factoid. This process is an essential part of becoming-animal, described by Deleuze and Guattari in the parable of the orchid and the wasp, where the former forms an image of the latter capable of deceiving the male wasp into spreading the flower’s pollen (Deleuze and Guattari Reference Deleuze and Guattari2013: 9, 278, 342). It is possible to interpret the ‘image’ in these two cases as a form of fiction that creates the conditions for a multispecies entanglement to enfold. Could we infer that the same is possible with sonic fiction?

This process is at work throughout the Cosmicomics, triggered by the unpronounceable name of the main character: Qfwfq. The SF novelist and critic Ursula K. Le Guin, unexpectedlyFootnote ¹⁴ comments that ‘one fault in [Calvino’s] prose is its joky or satirical convention of unpronounceable names … reducing [language] to the literally unspeakable symbology of mathematics’ (Le Guin Reference Le Guin2016: 161; emphasis added). Furthermore, she considers it as a flaw when ‘that [language] game gets chancy’ in the way that the reader articulates the sound and the text (ibid.; emphasis added). This gesture and the aesthetic judgement implied here can be read as an example of the anthropological machine: it appears caged in the binary of legibility/muteness, turning away from the zone of indiscernibility that threatens to disassemble the linguistic structure into creative non-sense, while it is this non that allows the reading body to momentarily escape the anthropological machine (Cimatti Reference Cimatti2020: 177). Thinking back to the work of George Lewis which was discussed earlier (Reference Lewis2002: 99), approaching the phonetically provocative name as ‘chancy’ recalls how ‘aleatory and indeterminism’ are words and ideas often used to bypass the word ‘improvisation’, by defining the possibility of other logics in terms of a lack (e.g., of structure, of sound, of meaning). On the contrary, the unpronounceable name amplifies the interactivity of reading and the role of the reader as an improviser. For that reason, unpronounceable writing was central to the notation of The Spiral. This will be discussed in section 6, after a deviation through another molluscan figure.

5. MEMORIES OF A VAMPYROTEUTHIS

Influenced by the ‘cultural revolution’ of the ‘Information Age’ in the 1980s in the wake of AI and genetic technology (Flusser and Bec Reference Flusser and Bec2012: 66, 74), Vilém Flusser (1920–1991) started writing a fable about the vampire squid (vampyroteuthis infernalis) in the form of an unusual scientific report, partly ‘written by’ vampyroteuthis infernalis itself and illustrated by the biologist, zoosystematician and artist Louis Bec (1936–2018). Lured by the paradoxical relationship of the vampire squid and the human, Flusser studied the creature as our existential ‘antipode’ (Flusser Reference Flusser2011: 36) attempting ‘to critique our vertebrate existence from the molluscan point of view’ (ibid.: 27). Vampyroteuthis Infernalis was the first work of what he saw as a new discipline of ‘Philosophy of Fantasy’ (ibid.: 146). He describes part of his research methodology in a letter he wrote to Dora Ferreira da Silva:

By inhabiting a molluscan perspective, the author intended to produce a new mode of philosophical writing that is ‘simultaneously scientifically exact and mad fantasy’ (ibid: 142). More recently, it has been adopted in posthumanist discourse as an epistemologically potent ‘media fable’, capable of questioning the ‘universally applicable’ human perspective, enacting a mode of ‘self-critique’ of the author’s anthroposition (Jue Reference Jue2020). In this way, it can inform the way we might approach more-than-human ways of listening. Even though in Vampyroteuthis Infernalis Flusser does not elaborate on sonic media, he was interested in the ‘vampyroteuthian’ turn in the development of artificial intelligence and the shift towards software-based art, claiming that ‘there can be no doubt that “soft” [in software processing] alludes to molluscs (“soft animals”)’ (Flusser and Bec Reference Flusser and Bec2012: 67). In this sense, he hints towards a potential space of (im)possibility, or sonic fiction, where machine listening was always already molluscan.Footnote ¹⁵

Machine listening is intimately linked with a machine of subjectivity, capable of rendering inoperative the anthropological machine of listening. In More Brilliant than the Sun: Adventures in Sonic Fiction, Kodwo Eshun develops a theory of sonic fiction emerging from the ‘intersections of sound and science fiction’ in the ‘coevolution of machines and humans in the late 20th C Black Atlantic Futurism’ (Eshun Reference Eshun1998: 00[-002]). It operates by transversal ‘sensation transference’ where ‘concept feeds back into sensation, acting as a subjectivity engine, a machine of subjectivity that peoples the world with audio hallucinations’ (ibid.: 7[121]). In sonic fiction, the boundaries of the human are often blurry, with various escape routes to animality. For instance, Eshun considers how the vocoder ‘[turns] the voice into a synthetic spectrum of perverse voco-imps [letting] you talk with cartoons, become cartoon, become animal, become supercomputer’ (ibid.: 06[080]). This idea could be traced back to John C. Lilly’s s(f)cientific experiments during the Cold War, who attempted to study human–alien communication through cetaceans ‘using the vocoders to translate dolphin vocalisation into human speech’ (Gerspacher Reference Gerspacher2022: 50). This ambition still resonates today, although the vocoder is replaced by more complex listening assemblages, involving robotics, linguistic models and machine learning (Andreas et al. Reference Andreas, Beguš, Bronstein, Diamant, Delaney and Gero2022). Even though Eshun does not directly consider sonic fiction in relation to machine listening, we could view the work of George Lewis as the articulation of the two, especially in his discussions on the notions of emotional transduction in relation to Eshun’s sensation transference mentioned earlier, and technology-mediated animism in relation to Eshun’s notion of machine life.Footnote ¹⁶

If sonic fiction is a subjectivity engine (Eshun Reference Eshun1998: 02[014], 03[034], 07[121], A[186]), what sonic engine could produce the sonic fiction of a molluscan subjectivity? Applying Flusser’s molluscan dialectics in the interaction design of the music composition, it might be possible to conceive of machine listening as an (anim)algorithm enabling this task. Flusser describes his method of SF speculation, which I refer to as (malaco)logic, in a letter to his friend, the engineer and philosopher Milton Vargas:

The syllogisms of Vampyroteuthis Infernalis gain momentum through this oscillation between analogues or homologues of the human/vampyroteuthis pair. For instance, Flusser’s theory of ‘vampyroteuthian art’ unfolds like a spiral around the analogy of ink, gradually amplifying the differences between the human use of ink for writing as objective inscription and the vampyroteuthian use of ink for hiding as subjective deception (Flusser Reference Flusser2011: 105–16). Thus, he goes as far as to define an ‘intersubjective society’ as ‘a society of Vampyroteuthes’ (ibid.: 114). In order to transcode this kind of molluscan subjectivity into a sonic engine, it might be suitable to apply Flusser’s (malaco)logics to the model of animal subjectivity developed by Jacob von Uexküll, now recognised as the pioneer of biosemiotics.Footnote ¹⁷ Contrary to most of the traditional conceptions of the animal, Uexküll approached it as ‘a living being with its own subjectivity’ (Cimatti Reference Cimatti2020: 28). He developed the theory of the umwelt, or the animal’s perceptual life-world (Uexküll Reference Uexküll2010: 1–34), which considers animal subjectivity as a closed unit consisting of the organism’s ‘perceptual world’ and ‘effect world’, where the former triggers a series of reactions or ‘adaptive behaviours’ that make up the latter (Cimatti Reference Cimatti2020: 29). Illustrated in the diagram of the ‘functional cycle’ (Uexküll Reference Uexküll2010: 49), the resonance between the two ‘worlds’ is what produces the behavioural complexity of living beings according to the theory of umwelt.

The machinic umwelt encountered by the performer of The Spiral is an example of (malaco)logic applied in machine listening: a neural network enables the resonance between the human/molluscan sonic analogy of a molluscan perceptual world and a human’s musical effect world. The former was modelled on recent biological studies of molluscan hearing. The data produced through lab experimentation in a study of the hearing of the bivalve mollusc Magallana Gigas (Charifi et al. Reference Charifi, Sow, Ciret, Benomar and Massabuau2017) was used to emulate the intensity and delay of the oyster’s capacity to be affected by a spectrum of acoustic vibrations through a patchwork of digital filters and delays. The processed sound is then quantised, normalised and used as the input for machine learning. Since the oysters proved to be highly sensitive to frequencies between 10 and 80 Hz, a geophoneFootnote ¹⁸ is used instead of a conventional microphone, as it is more sensitive to the low register and the infrasound that affect the molluscan body most.

The machine ‘listens’ through a fuzzy Adaptive Resonance Theory (ART) model for unsupervised machine learning, producing a flux of constantly updating clusters of identifiable percepts. The version used is implemented into a Max objectFootnote ¹⁹ by Benjamin D. Smith, who envisioned its potential to be used ‘as an intelligent partner, listening to the musical development and making informed choices and mappings as the improvisation unfolds’ (Smith and Garnett Reference Smith and Garnett2012). The Spiral works in a similar way. Every time a new cluster is detected, the percept that triggered it is associated with a set of data making up the ‘effect world’, consisting of parameters for sound synthesis that are pre-composed as ‘optional’ material to be used throughout the piece (that are almost never reproduced accurately). The association is done using another neural network, a multilayered perceptron extending the clustering process into a continuous micro-temporal capacity for reactive behaviour. This behaviour is expressed through ‘musical’ sound by modulating the granulation parameters that re-synthesise the live input of the human performer into radically different and often surprising gestures. This hybrid umwelt – half human, half molluscan – makes up an (anim)algorithm with a functional cycle that opens a zone of indistinction for becoming-animal to unfold.

6. MEMORIES OF AN AURIFORM

The unpronounceable notation of The Spiral was attained through a (malaco)logic of hearing emerging around the notion of the auriform.Footnote ²⁰ This malacological term, which is used to classify molluscs ‘having the shape and form of a human ear’, sets up a Flusserian dialectic oscillation between the human hearing apparatus and the molluscan image. Inverting the terms of the analogy, instead of the auriform molluscs having the shape and form of a human ear, one could look at the human ear as having the shape and form of a mollusc. Subsequently, the plurality of molluscan visual forms can be viewed as a space of (im)possibility for imagining (post)human ears and ways of hearing.

In this sense, the method of notation explored in The Spiral engages with the sonic fictions of our molliform earsFootnote ²¹ by tracing forms of molluscs on systems of musical staff. This technique draws on the work of eco-acoustic composer Matthew Burtner. Part of his composition Cloudprints (Reference Burtner2007) is composed by digitally superimposing a linear grid system on pictures of clouds to sonify their intricate morphology. However, in The Spiral, the linear grid is substituted by groups of concentric circles that, according to Calvino and Flusser, would be more adequate for accessing a molluscan existence. For instance, Flusser infers that ‘whereas we think linearly (“rightly”), [the vampyroteuthis] thinks circularly (“eccentrically”)’ (Flusser and Bec Reference Flusser and Bec2012: 42). The theme of circularity reappears consistently in molluscan SF, which Qfwfq explains in terms of radial symmetry:

It is worth mentioning how this circularity refrain echoes in the SF film, Arrival (Villeneuve Reference Villeneuve2016).Footnote ²² Appearing as the foundation of the cephalopod-like aliens’ tentacular existence, circularity is visually illustrated through the language of these creatures, in which they ‘write’ by manipulating an ink-like substance into complex, circular floating formations. As the narrative unfolds, we understand that the circularity of this language determines the temporal structure of the writing subjectivity, which eventually explains the time-warping visions of the protagonist, Dr Louise Banks. In his book The Owls Are Not What They Seem: Artist as Ethologist, Arnaud Gerspacher interprets the film in tandem with Derrida’s parable of the cuttlefish, where the mollusc’s ink ‘would have the power of attesting to cephalopod interiority’, or the ability to be or do (prior to saying) ‘I’ (Reference Gerspacher2022: 48). For him, both stories are illustrations of what he calls the ‘ethogrammatological reversal’ – a hypothesis that questions the exceptionalism traditionally attributed to human language, considering it to be only one communicative medium among a plurality of semantic possibilities grounded on animal behaviour (ibid.: 45). By following this hypothesis from mythology to ethology, one might indeed extend a meaningful (musical) ‘I’ beyond anthropocentrism through the study of animal behaviour and trace structures.

Contrary to the popular imaginary, molluscs do not have radial symmetry, but bilateral symmetry like humans (Flusser and Bec Reference Flusser and Bec2012: 12). However, Flusser points out that they ‘resign half of their body in favour of the other to become “half animals”’, or ‘ek-centric animals whose bodies incline towards coiling both as a whole and in all their details’ (ibid.: 12–14). This ambiguity between bilaterality and radiality might explain his observation that the ‘spiral symmetry is the fundamental theme of the mollusc body’ (ibid.: 14), a theme that is also visible in the image used for the second section of the composition (Figure 1). The photograph captures the trace of a nudibranch mollusc, signifying the mollusc’s absence within the presence of the spiral articulation of its eggs. During the performance, the spiral trace re-emerges in the zone of indiscernibility between the circular writing informed by the molluscan morphology and the linear reading typically imposed by Eurocentric music culture.

Figure 1.

Nudibranch eggs in a spiral arrangement. This image was used in the composition of the second movement of The Spiral, titled ‘Ova Nudibranchiorum’. Photo: Alexandros Patakis.

The Spiral consists of seven sections, each based on a picture of a different mollusc, or other aquatic creatures found in a molluscan milieu.Footnote ²³ The morphology of each creature was then inscribed on the circular system using conventional Western music notation, which is warped into unreadable formations at several points of ‘dissonance’ between the linear languages and the circular structure. The performer is asked to play each of the seven sections starting from the inner circular system, gradually proceeding outwards ad lib before transitioning between circles. At each circle, the gaze of the performer starts at the imploded graphics at the circle’s centre before spiralling outwards. The circularity of the inscription method pushed the score to the limits of legibility from a human point of view. It demands the constant spiralling of vision while it occasionally warps the familiar symbols to extremes (Figure 2). In this way, the (malaco)logical traces turn the composition into an unpronounceable improvisation unfolding as the auriform music meets the plurality of molliform ears.

Figure 2.

The third movement of The Spiral, where the legibility of the notation collapses as it approaches the centre of the circular structure.