314
Luciana Parisi / Recursive Philosophy and Negative Machines
This article addresses these tendencies and argues that a systematic approach to nature and technology also needs to account for the problem of
the “philosophical decision structure” in mediatic thinking, instrumentality,
and automation.3 In what follows, the pillars of this structure will be explored
through the cybernetically influenced notions of recursivity, self-organization,
and self-programming. As it may become clearer later, if decision-making
is already given in the world as a natural evolution of human reason, it is
because its structure is sustained by the flesh of the medium. To explore the
challenges that mediatic thinking, instrumentality, and automation pose to
the decisional structure of philosophy, I draw some insights from Get Out
(dir. Jordan Peele, 2017).
Placed among horror, fiction, and politics, Get Out offers a contemporary
map of how the thin line between automation and philosophy, namely
servomechanic intelligence and transcendental reason, reminds us that
the question of technology cannot be separated from the violence of colonial
slavery.
The movie opens with an unsettling scene of the abduction of a young
black man, choked and shoved into the trunk of a car in a quiet suburban
neighborhood. In the next scene, Chris, a black photographer, and Rose,
his white girlfriend, plan to go out of town to meet her parents. Chris, however, doesn’t leave unguarded; his automatic camera is always strapped
around his body. Chris and Rose are welcomed by professionals of the mind;
her dad is a renowned neurosurgeon and her mom a progressive psychotherapist. The plot thickens when Chris is unwillingly hypnotized whilst having a
cup of tea. He is sent to the noumenal space that lies beneath the world of
transcendental philosophy. He traverses the phenomenal wall into a vortex
of dark optics, on the other side of the screen of consciousness. Chris’s mind
enters a phase of deprogramming to a degree zero; the biophysical ground
that secures his understanding of the world rolls down into a black abyss.
Here, thinking coincides with no decision, no expression, no action. All that
is left of Chris is the unstoppable thought processing that engulfs consciousness into unknowns, single sequences of unlimited images. When Chris is
awakened, nothing feels the same. He now exists in a twilight zone between
intelligent automation and transcendental reasoning.
3. François Laruelle, Principles of Non-Philosophy, trans. Nicola Rubczak and Anthony Paul
Smith (New York, 2013), p. 231.
L u c i a na Pa r i s i is professor in the Program in Literature and is core faculty
for the Computational Media Art and Culture Graduate Program at Duke
University.
Critical Inquiry / Winter 2022
The Armitages’ master plan of preserving the white brain of their descendants in the slave machine of black bodies is called the Order of the Coagula.
This transhuman enhancement relies on the servomechanic automation of
black bodies, a medium of thought affording the colonial abstraction of value
in the slave machine. What Chris is after, however, is not simply the unveiling of the master-slave dialectic. Instead, the negative optics of his camera—
where flashes coincide with autoimages of darkness—allow for an automation of another kind. Here, technothinking, instead of defining the end of
philosophy and the crisis of reason, shows that black automation cannot
be subsumed to a priori decision, whether this implies a biological or mathematical axiom.
What follows is a nonphilosophical envisioning of automation inspired
by François Laruelle’s focus on the dark optics of nonphotography, a negative negation in the thinking medium.4 I argue that the negative negation of
the slave-machine takes both philosophy and automation beyond the light
of decision and within nonoptical darkness. Here, philosophy and automation are neither coupled nor set in opposition to each other. Instead, philosophy and automation are asked to refuse their self-determining axiomatics
and host the heretic optics of negative machines. As much as philosophy can
no longer hold onto the authority of decision-making, so too computation—
as the automation of reasoning per excellence—can no longer grant the extension of human cognition. The question of philosophy after computation
requires a critique that unlocks the negative side of the decisional structure
of philosophy, namely mediatic indeterminacies.
From this standpoint, it is important to consider how automation in the
form of computational processing has itself changed in the period between
post–World War II to today. From the computational application of symbolic logic (between the 1950s and 1970s) to the advance of statistical probabilities (between the 1980s and 2010s) and most recently of artificial neural networks (from the early 2000s to today), computational modeling no
longer matches given concepts to objects. One can map the shift from symbolic logic to learning algorithms in machine translation for instance, a subfield of computational linguistics: from the ’50s Rule Based Machine Translation (RBMT) to the late ’80s Statistical Machine Translation (SMT) and,
most recently, the ’00s Neural Machine Translation.5 If RBMT programmed
translation as the pairing of the linguistic structure of the input language with
4. See Laruelle, Le Concept de Non-Photographie/The Concept of Non-Photography, trans.
Robin Mackay (New York, 2015); hereafter abbreviated L.
5. See W. John Hutchins, “Machine Translation: A Brief History,” in Concise History of the
Language Sciences: From the Sumerians to the Cognitivists, ed. E. F. K. Koerner and R. E. Asher
(New York, 1995), pp. 431–44, and Dzmitry Bahdanau, Kyunghyun Cho, and Yoshua Bengio,
315
316
Luciana Parisi / Recursive Philosophy and Negative Machines
that of the output language, starting from grammatical rules and semantic
meaning, SMT used word-by-word alignment probabilities based on the position’s alignment of words in order to obtain a maximum likelihood result.
However, with statistical translation, automation already entered the realm
of probability calculations, whereby results, despite being predictable, are not
given from the start.
With syntax-based statistical translations, the source-language analysis
of a sentence made of parts was replaced by a target-language string, involving stochastic operations at each node. In 2013, this statistical alignment
model was set aside in favor of a new end-to-end encoder-decoder structure,
which introduced a class of probabilistic translation models called Recurrent Continuous Translation Models (that is, continuous representations
for words, phrases, and sentences). Today’s Deep Neural Networks, which are
applied to sequence-to-sequence learning, show that neural machine translation is built on a single large network trained to read and translate a sentence.
Without having to program parts of a sentence as hard (fixed) segments,
neural machine translation automatically looks for parts of a search sentence
that are relevant. As a result, the prediction of a target word depends on the
contextual parts of the sentence. In other words, whereas Phrase-Based Machine Translation (PBMT) breaks an input sentence into words and phrases
translated largely independently, Neural Machine Translation (NMT) learns
from the contextual use of an entire input sentence, which becomes a unit for
translation.
These examples show the transformation of the automation of decisionmaking into a strategy of prediction and control based on learning and feedback. With computation, the modern pillars of transcendental reasoning
have shifted from self-determining truths that fit all proofs to the feedback
activities of learning algorithms. This also entails a shift from a universal
mathematics of truths applied to objects to an interactive logic of rules whose
syntactical arrangements come to include undecidable states that can only
define truths retroductively, that is, after the interaction between rules has
taken place.6
Arguing that the question of technology cannot be separated from the violence of colonial slavery, therefore, also means that philosophy after computation entails a global transformation of automated reasoning in terms
of cybernetic learning. This transformation implies the principle of recursivity—a feedback function that reorients the teleological purpose of a system
“Neural Machine Translation by Jointly Learning to Align and Translate,” paper presented at
ICLR 2015, San Diego, May 2015, arxiv.org/pdf/1409.0473
6. See Jean-Yves Girard, “Locus Solum: From the Rules of Logic to the Logic of Rules,”
Mathematical Structures in Computer Science 11 (June 2001): 301–506.
Critical Inquiry / Winter 2022
towards local iterations. Instead of a vulgar mechanization of philosophy or a replacement of conceptual categories with automated tasks, recursivity addresses
systemic knowledge in terms of temporal change, showing that incompleteness
constitutes both philosophical and automated systems (see R).
However, as it may become clearer later, it may be insufficient to recuperate recursivity as a metaphysical principle that aligns philosophy and automation. Instead, the emphasis on recursivity may oversee the problem of decision as formal and efficient causation, universal and particular truths, that
define the structure of philosophy. Recursivity may instead only allow that
self-decisions become preserved in automated self-reflection. This extension of the decisional structure of philosophy within machines precludes
the possibility of a machine philosophy without master, of algorithms without programming, and of automated intelligence without the slave machine.
The plan to extend self-decision returns in the plot of Get Out, where
recursivity transplants transcendental philosophy in the slave machine. A
recursive principle of knowledge is at play here. The servomechanic medium becomes the flesh for colonial abstraction in the project of prosthetic
automations. Get Out, however, also exposes the reticular relation between
automation, media, and blackness as an opportunity for rethinking technology away from the pillars of transcendental metaphysics, recursivity, and
autopoiesis.
One needs to return to the argument of undecidability in computation
to discuss the reticulation of automation, medium, and blackness in terms of
a nongiven metaphysics. More importantly, undecidability is not opposed to
determination but rather becomes the possibility for nonphilosophy as a
negative machine.7 Here, philosophy is conditioned by what the instruments
refuse to reproduce, while the real enters computation. Philosophy after computation, therefore, can only start as a quest for the outside, namely what has
always already been expulsed from the decisional structure of philosophy.
This is not absolute contingency where machines capture the real without
mediation. Instead, the externalities of decision take the status of iterative
functions in computation to demarcate the algorithmic practices of negative
mediation. This negative negation takes the slave machine as corresponding
not only to the exclusion of the medium from philosophy but, more importantly, to the negative darkness of automated machines. Through Get Out, in
particular, one can discuss this negative darkness in terms of the negative decidability of the means—namely the way slave machines refuse to reproduce
transcendental categories and instead pave the way to get out from the recursive extension of transcendental decision in machines.
7. See Laruelle, Principles of Non-Philosophy.
317
318
Luciana Parisi / Recursive Philosophy and Negative Machines
This argument for negative machines, however, necessitates first that we
approach automation beyond the Heideggerian critique of the modern
question of technology.
Techne Ontology
In the aftermath of World War II, Martin Heidegger issued a warning
against the demise of philosophy at the hands of cybernetics. He claimed that
cybernetics brings philosophy to its own completion by dissolving itself into
a general science of information. What he called “the question concerning
technology” imparted a new order of causality, or autocausality, onto human
culture.8 In particular, industrial capital had created a new field of existence
where the causality of decision had become opaque. This opaqueness manifested in the modern form of techne, in the servomechanic system of accumulation. By lamenting the impossibility of redirecting means towards ends
beyond the industrial abstraction of value, Heidegger argues that the “bringingforth of the true into the beautiful was called technē.”9 Techne here indicates
the need to go back to a time when techne was one with poesis—or crafting—before the automation of reason.
But isn’t this warning against the end of philosophy precisely a way to preserve the metaphysics of self-decision by lamenting that means now have
ends that no longer correspond to the transcendental program of thought?
Doesn’t cybernetics precisely defy this deductive assumption of a universal
and formal structure of thought?
If the cybernetic age eventuates the loss of being, which technology both
reveals and obscures, it also demarcates a rupture within recursive epistemologies in the form of instrumentalities that no longer obey given ends. From
this standpoint, Heidegger’s vindication for poiēsis (as the lived experience
or the temporality of the artifact) against technology does not allow one to
envision an improper becoming of automated knowledge. As intelligent
machines challenge the field of transcendental decision, they also construct
“philo-fictions,” stemming from the negative stance of a nondecisional thinking.10 Philo-fiction, according to Laruelle, is parallel to science fiction in its
capacity to refuse axiomatics and clone the real as unrepeatable singularities,
unmatched to given representations.11
8. See Heidegger, “The Question Concerning Technology.”
9. Ibid., p. 34.
10. See Laruelle, The Last Humanity: The New Ecological Science, trans. Anthony Paul Smith
(New York, 2020).
11. See ibid.
Critical Inquiry / Winter 2022
Philo-fiction can first confute the assumption that machine intelligence
comes after transcendental philosophy and similarly that technology only
serves to unmask the limits of Western metaphysics and capital’s abstraction. As Haraway famously pointed out, cybernetics’ circuits of communication have not simply delivered secure ends for the good of humanity but
have instead exposed philo-fictional possibilities against the master narrative of the human.12 Central to this article is the argument that the Darwinian model of evolutionary speciation, based on the racialization and gendering of the survival of the fittest, was already challenged by philo-fictional
thinking in machines: namely, the artificial intelligence of servomechanic
intelligence. The negative machines of capital’s abstraction rather expose
the philo-fictional possibilities of instrumentality as an improper presence
of inhuman thinking in nondecisional and noneffective patterning.
Negative Machines
Perhaps one can begin by suggesting that the tension between reason and
intelligence is intrinsic to the modern project of philosophy. The argument
for a scientific method of and for philosophy, also called “pure reason,” returns in the epistemological quest of cybernetics—namely, intelligence.13
Recent debates, however, seem to assume that before the arrival of automation, philosophy was a pristine form of knowledge, immune from the efficient causality of techne. That philosophy must be recuperated away from
the worldview of technocapital also implies that one must hold on to a fundamental bifurcation in the decisional structure of philosophy, whereby automation stands as the negative negation of reason.
As automation is placed on the dark edge of philosophy, it is either said
to side with the negative term of exclusion, a vacant servomechanic tool, or
with dominant power, a means to extend capital’s violence and state governance (by ideological, discursive, affective means). In discussing this binarism, this article aims to question first the argument that machines, the
medium, the instrument of thought are excluded from the authoritative
(or decisional) autonomy of philosophy and from the postulates of pure
reason. It then follows the view endorsed by the post-Kantian critique of
the rational subject where automation is precisely evoked to challenge
12. See Donna J. Haraway, Simians, Cyborgs, and Women: The Reinvention of Nature (New
York, 1991). My use of philo-fiction here pushes Laruelle’s vision of nonphilosophical decision
to argue that the onto-epistemological pillars that define the human in terms of self-posed
decision or efficient decision in automation must be challenged.
13. See Immanuel Kant, Critique of Pure Reason, trans. and ed. Paul Guyer and Allen W. Wood
(New York, 1998).
319
320
Luciana Parisi / Recursive Philosophy and Negative Machines
transcendence.14 The typewriter, moving images, and computational systems
have been enduring instances of the capacity of automation (or technology
in general) to defy the authority of rational thought. From the notion of the
optical unconscious to the orality of thinking, sensible automata, and nonconscious cognition, automation has been used to highlight the limits of
transcendental philosophy. As the negative of pure reason, however, automation has also been taken to act as a placeholder to demonstrate how capital, patriarchy, and colonialism hide power beyond the naturalization of
technology in the form of a prosthetic extension of humanity. As having
no soul, except the less-than-human soul of an empty vessel excluded from
philosophy, automation is said to bring together humans, animals, and machines. This post-Kantian argument for automation contains within itself
the implication of an ontological originary coconstitution of thought and
machine.
However, to speak with Denise Ferreira da Silva, one can argue that this
duplicity of philosophy and automation (reason and intelligence) is yet another manifestation of what she calls “the transparency thesis.”15 With this
notion, da Silva refers to how the universality of the subject lies behind the
constant reproduction of racial violence and antiblackness, which returns on
bodies through the category of difference. As much as transcendental reason
prescribes how the self-determining subject defines difference through the
lenses of the universality of history and analytics, so too the transparency thesis takes the other or the object as difference within the structure of the given,
for which the subject, the self, and the concept are predetermined as one. In
particular, the transparency thesis founds the ontology of the self-determining
subject with the modern secularization of knowledge, ensuring the persistence of an “ontoepistemological account that institutes ‘being and meaning’
as effects of interiority and temporality” (T, p. 4). According to da Silva, this
account relies on the modern historical and analytic forms that perpetuate
the essence of being through the explanations that have granted universality
to the originary decision or self-determination of modern philosophy.
The question posed by da Silva is a radical one: How do we reinvent postKantian critique without reproducing the mirroring trick of the transparency thesis? In other words, how do we challenge the perpetuation of selfdeterminate knowledge without reproducing its violence? Is it sufficient to
14. See Jacques Derrida, “Typewriter Ribbon: Limited Inc (2) (‘Within Such Limits’),” trans.
Peggy Kamuf, in Material Events: Paul de Man and the Afterlife of Theory, ed. Tom Cohen et al.
(Minneapolis, 2001), pp. 277–360, and Gilles Deleuze, Cinema 2: The Time-Image, trans. Hugh
Tomlinson and Robert Galeta (Minneapolis, 1989).
15. Denise Ferreira da Silva, Toward a Global Idea of Race (Minneapolis, 2007), p. 1; hereafter
abbreviated T.
Critical Inquiry / Winter 2022
unmask the universal game of inclusion/exclusion without remaining entrapped in the account of the already historically constituted and excluded
others?
Da Silva’s incisive critique of the inclusion/exclusion dyad is not addressed towards automation or the thinking medium of philosophy; nonetheless, it is central to this article’s argument. In particular, one must point
out that da Silva’s warning against the perpetuation of the transparency
thesis shows the racialized and gendered origination of modern philosophy. The latter entails the unity of the historical and analytic schema based
on interior loops of time from which the bifurcation of the included/excluded coupling continues to exercise colonial violence in the space of exteriority. Similarly, da Silva suggests that the work of critique cannot repeat
this old trick of modern metaphysics—namely, in vindicating the name of
the other, critique seems to persistently contribute to expanding the transparent subject under the vestiges of a sociohistorical objective analysis of
the logic of exclusion. Instead, she warns us against the cul-de-sac in which
critical theory has found itself in remaining analytically attached to the
double pincer of ontoepistemology, without defying its originary structure.
Instead of focusing on expanding the capacities of including more of the
excluded in transcendental self-determination, da Silva denounces the way
this view sets in place the impossibility of thinking beyond the violent, gratuitous perpetuation of antiblack onto-epistemology.
Da Silva’s contribution to this current discussion shows how the binarism
between intelligence and reason mainly appears as an onto-epistemological
strategy of othering (or inclusion/exclusion) rooted in the servomechanic
principles of transcendental philosophy and cybernetics. It follows that the
perpetuation of the transparency thesis also relies on the coconstitution of
philosophy and automation insofar as these are both part of the larger schema
that explains (and justifies) the transcendental origination of reason as a
decisional structure of thinking. As opposed to the transparency thesis, da
Silva instead insists upon the spatial exteriority of an infinity that cannot
be resolved in the binary logic of yes and no. The critical analysis of the other
in terms of transcendental self-determination cannot account for a radical
exteriority but will inevitably succeed in reconducting the other to already
“differentially constituted historical beings before their entrance into the
modern political spaces” (T, p. 5). In addition, da Silva goes as far as to
say that the rhetoric of the critique of the excluded leaves the neglected negative side of transcendental reason in “a self-defeating kind of transparency”
(T, p. 5). This rhetoric of the excluded ends up preserving self-determining
reason through and with the medium of the other for which no future is
possible.
321
322
Luciana Parisi / Recursive Philosophy and Negative Machines
However, da Silva’s conception of a global idea of race maintains that
there is no universal modern philosophy—and thus no sustainable appeal
to pure reason—without material exteriority (or material infinities, what
she elsewhere calls “0 [blackness]”) from which to hijack the schema of
inclusion/exclusion.16 Borrowing from these reflections, this article attempts
to offer a perspective of exteriority on automation without transcendental
interiority. But what exactly would this argument for material externality
and infinite fractalities mean for our discussion about philosophy after
computation?
If the practice of critique has worked to lift the curtain of philosophy as
a form of a priori decision-making always already operating through exclusion, this article envisions the need for both philosophy and automation
to flex their axiomatics toward the exteriority of negative propositions,
or heretic thinking in machines. The goal here is not to repair philosophy
through or via machine intelligence but instead to carry out the argument
for an asymmetric (fractal and nonrecursive) tension between reason and
intelligence. The question of technology (and of the limits of transcendental
philosophy) can be readdressed in terms of a retroductive inquiry into the
modern formation of a generic artificial cognition—whereby the colonial
entanglement with technology shows that the binary intelligence (or learning) versus reason (or reflective judgment) is each and every time conditioned by indeterminate externalities (or incomputables).
Instead of a monologic schema that must be constantly unveiled and
destructured, philosophy after computation becomes a stereo-logic philofiction of worlds “without separability.”17 Da Silva discusses the structural
imperative of separability at the core of the modern text as the juridical,
spatial, and temporal parts that constitute the world order according to
particular epistemologies of measurement. The Kantian modern project relies on separability, namely the way the limit of reason defines what can be
ordered by spatiotemporal intuition and the categories of understanding
(see “O,” p. 60). Instead, she proposes to turn to postclassical physics, relativity and quantum mechanics to retheorize thinking tools—methods and
approaches—that would not reproduce separability and sequentiality. Her
reflections concern the need to transform the epistemological conditions of
16. Da Silva, “1 (Life) 0 (Blackness) = ∞ − ∞ or ∞ / ∞: On Matter beyond the Equation
of Value,” e-flux journal 79 (Feb. 2017): www.e-flux.com/journal/79/94686/1-life-0-blackness-or
-on-matter-beyond-the-equation-of-value/
17. See Da Silva, “On Difference without Separability,” in Incerteza Viva (exhibition catalogue, 32nd Bienal de São Paulo, São Paulo, Brazil, 7 Sept.–11 Dec. 2016), pp. 57–65, issuu.com
/bienal/docs/32bsp-catalogo-web-en; hereafter abbreviated “O.”
Critical Inquiry / Winter 2022
knowledge and argue against the “limited picture of The World” structured
by the model programs of Newtonian and Einsteinian physics where separability becomes an ontological principle that reproduces sociocultural,
existential, legal, and political understanding of difference and of the human (“O,” p. 64).
As instances of scientific (and technoscientific) models that were set to
extend the rational mission of transcendental philosophy, cybernetics and
computation are here approached as speculative instrumentalities, questioning the conditions of knowledge. For instance, it will be suggested that
computation and cybernetics have importantly transformed these conditions, no longer deduced by transcendental categories or simply by the empiricism of data associations. Instead, nondecisional practices in automated
intelligence can be understood as incomputables—or indeterminate complexity in sequential patterns. To defy the axiomatic method of a priori decision, therefore, it is important that we study artificial intelligence in terms
of the philo-fictional mediations of indeterminacies.
If, against teleological metaphysics, machine thinking can provide a method
of knowledge that challenges the a priori intuition of space and time, the computational analytics of machine intelligence can in turn expose the materiality
of nonlinear functions as external fractalities of space and time.
Recursive Metaphysics
According to Norbert Wiener, the automatic machine is not simply a tool
that repeats specialized functions.18 For Wiener, this assumption perpetuates
the social supremacy of the master-slave dialectic, whereby the automatic
machine remains contained by the norms for which machines like slaves
only respond to orders. This duality sees machines as included/excluded and
thus either as obeying slaves or threatening rebels.19 Wiener instead proposes
a theory of learning that could be generic enough not to be reducible to sheer
mechanism. By taking feedback as a central function that modifies behavioral
patterns, he defines intelligence.
What cybernetics rejects from transcendental philosophy is precisely the
assumption that the conditions for thinking are universally given in the spatiotemporal coordinates of the human. Instead, for cybernetics the temporality of feedback shows that what appears as a passive servomechanic function
of repetition transforms the conditions for learning—namely, allowing a
18. See Norbert Wiener, The Human Use of Human Beings: Cybernetics and Society (Boston,
1954).
19. See Louis Chude-Sokei, The Sound of Culture: Diaspora and Black Technopoetics
(Middletown, Conn., 2016).
323
324
Luciana Parisi / Recursive Philosophy and Negative Machines
machine to act in the future.20 Here, space and time are not structured intuitions but depend upon prediction. Machine intelligence indeed corresponds
to the anticipation and modification of future behavioral patterns.
It may be useful to pause here. Perhaps one could play devil’s advocate
and ask: What does intelligence have to do with the transcendental judgment after all? We know that intelligence implies a modification in patterns
of behavior, but reason instead commits to postulates that can be explained
according to a logical procedure granted by verification or proof—a nondogmatic reflectivity that grounds knowledge. Instead, the servomechanic
function of feedback has a minimal, if any, basis for thinking about thinking.
As opposed to the view about the cybernetic downgrading of reason into automated intelligences,21 the skeptic may still want to hold on to the distinction between intelligence and reason, as the latter may still allow for a recuperation of critical thinking today.22
As Yuk Hui has recently suggested, however, the modern project of philosophy shows that the transcendental schema of reason is particularly entangled with the natural philosophy of the eighteenth century. Hui’s goal is
to explain how natural philosophy approached the problem of contingency
or indeterminacy through the articulation of recursivity, a fundamental principle of the organic. According to Hui, recursivity is central to reflective judgment because it keeps active the feedback between premises and results, ensuring adaptive response to the reality of contingency upon which the system
of rules comes to depend. The uncertainty of the external world becomes
part of the recursive loop of adaptability. Here it is not the given intuition
of space and time that ensures a universal condition of thinking and knowing
but the natural principle of recursive feedback that entangles the organic and
the inorganic in the continuous ordering of contingency. Transcendental interiority is thus derived from the time it takes for a loop to complete itself.
Time comes to define what links the gaps between points or the manner
in which points constitute recursive temporality.
For Hui, recursivity in natural philosophy demarcates a historical route
for both modern philosophy and cybernetics as models that confront the irregularities of nature. In particular, recursive self-reflectivity is embedded in
the epistemological vision of biology and the natural sciences of the eighteenth century that explained the world from the standpoint of the organism
(see R, pp. 63–64). Hui’s efforts to reconstruct the lineage of systematic
20. See Wiener, The Human Use of Human Beings, p. 71.
21. See Paul Erickson et al., How Reason Almost Lost Its Mind: The Strange Career of Cold
War Rationality (Chicago, 2013).
22. See Bernard Stiegler, The Future of Work, vol. 1 of Automatic Society, trans. Daniel Ross
(Malden, Mass., 2016).
Critical Inquiry / Winter 2022
thinking moves from natural philosophy’s conceptions of the organism to
the arguments for organicism and organology to offer an alternative history
of philosophy of technology, grounded not in mechanicism but in autopoiesis, self-organization, or recursivity (see R, pp. 42–43). As much as the
instability of nature or the problem of contingency enters the realm of systematic philosophy, so too a principle of regulation, defined by feedback,
grants ontological certainty to reflective thinking. As Hui claims, these
natural-philosophical permutations are constitutive of the history of technology (see R, p. 47). Hui also points out that “the attempt to know the Unknown without really knowing it . . . constitutes the spirit of organic thinking from Kant to cybernetics” (R, p. 229).
That natural philosophy provides the epistemological ground to explain
the common origin of modern philosophy and modern technology not only
challenges the dichotomy between vitalism and mechanicism but also contributes to questioning the assumption that cybernetics comes after reason.23
Hui’s analysis of the question of technology does not take cybernetics to announce, like Heidegger, the demise of philosophy and the withdrawal of being from the world. Instead, it entangles modern thought with natural philosophy and challenges modern metaphysics by explaining thinking in terms
of self-organization. One could argue that recursivity contains both reason
and intelligence insofar as these complementary and reciprocal activities
aim to unthread the knots of the unknown at the level of material cognition.
As much as reflective judgment relies on spatiotemporal intuitions to analyze
the patterning of the world, so too these adaptive (and learning) patterns
provide judgment with a schema of regularities stemming from the compression of external contingencies.
Following Hui’s project, it is possible to reject the argument about the
all-calculative origin of cybernetics (like Heidegger’s view of techne as the
enframing of the world under the universal logic of communication exchange)
and argue for a modern philosophy of technology historically entangled with
the philosophy of organisms. In particular, what Hui calls “cosmopoiesis,” inspired from Gilbert Simondon, points to a philosophy of technology as a genetic or ontogenetic process of knowing (R, p. 226). For Simondon, technical systems are not efficient executors of given instructions; instead, they are
conditioned by a “‘margin of indetermination’” or the world of contingency
(R, p. 189). Rather than providing a safe shelter from the unknown of nature, where technology coincides with calculative functions, the ontogenetic
process of knowing uncovers cybernetics as process or as in-formation. The
23. This is an argument that seems to be in contrast with the historical explanation of the
Cold War origin of cybernetics and automated decision-making brought forward by Lorraine
Daston, among others; see Erickson et al., How Reason Almost Lost Its Mind.
325
326
Luciana Parisi / Recursive Philosophy and Negative Machines
self-regulatory process of adaptation defines information in terms of an activity where reflective judgments are feedback loops that give shape to the
world. The margin of indetermination, however, continues to add randomness to the self-regulatory loops as much as the system faces contingencies that
force learning patterns to transform its initial premises. Hui’s project suggests
that cybernetics, far from ending philosophy or instigating the crisis of philosophy, entails an alternative metaphysical project (see R, p. 245). This critical
engagement with technology offers an account of the modern coconstitution
of organic-inorganic, human-machine, philosophy and automation.
Importantly for Hui, this systematic coexistence does not share the claims
of posthumanism or of transhumanism. While on the one hand, posthumanism offers anthropomorphic accounts of the machine-organism dynamics, transhumanism understands the organism in terms of individual
and programmable functions. Hui instead argues for cosmopoiesis, whereby
recursivity addresses the epochal transformations of the machine-organism
relation over time. Cosmopoiesis also accounts for the “remains of the inhuman” as the margins of contingency that are intrinsic to this relation
(R, p. 226). Hui takes inspiration from Jean-François Lyotard’s concept of
the inhuman as entailing two main viewpoints.24 On the one hand, the inhuman is what regimes of calculation or representation do to nature, the human, and reason—namely, as Max Horkheimer and Theodor Adorno lamented, reducing philosophy to the irrationality of technocapital. On the
other hand, the inhuman coincides with what cannot be known in advance,
a symptom of the incompleteness of being—namely, unprovable propositions in mathematical logic or incomputables in computational procedures.
The inhuman accounts for what remains outside the recursive feedback of
adaptability and yet forces the rules of the system to bend, transform, fail.
Hui’s insistence on the inhuman of systematic thinking also comfirms
that the coexistence of philosophy and cybernetics entails more than a prosthetic extension of thought in machines. The inhuman instead holds on to a
heretic position, a nonmutuality between philosophy and automation in the
expanded epistemology of nature. In particular, the inhuman comes to suspend originary organicism and the view that technology is anthropologically
universal (see R, p. 265). By closely following Simondon’s rearticulation of
cybernetics away from techne, function, mechanism, Hui proposes a theorization of cosmotechnics—cosmologies of machines—that cannot be contained in one universal system of measuring; science and technology are
bounded in a broader cosmic reality (see R, p. 39).
24. See Jean-François Lyotard, The Inhuman: Reflections on Time, trans. Geoffrey Bennington
and Rachel Bowlby (Stanford, Calif., 1991).
Critical Inquiry / Winter 2022
It may be useful to clarify here that this is not simply a matter of reclaiming a particular repressed cosmotechnics that has been excluded from the
anthropos of the earth. Instead, this insistence on the inhuman implies a resistance against the monologic of techne, as a “search of pluralism as indetermination, and therefore as multiple cosmotechnics” (R, p. 269). It is Hui’s
promise for a multiplicity at the core of cybernetic metaphysics that contributes to reapproaching the question of technology in terms of epistemological explorations of machine-organism relations. One can argue that this is
the starting point for questioning the view that computation comes after
philosophy.
However, as much as this promise opens new perspectives, it also prompts
us to wonder further about how to expand this critical intervention about
epistemologies of organisms and machines without reestablishing a given
myth or a decisional structure from where this relation is explained. By pushing this view further, one may ask: To what extent can the principle of autopoiesis be said to account for expanded epistemologies of cosmotechnics?
Would cosmotechnics, as a metaphysics that challenges anthropomorphism,
necessarily need to break from the biocentric principle of autopoietic adaptability? If natural philosophy can account for the coconstitution of modern
philosophy and technology in terms of an autopoietic structure of the living,
doesn’t this view of cosmotechnics also risk falling prey of biologism representing itself under the vestige of a current epochal techno-organology?
Before addressing these questions, it may be useful to clarify again why the
persistence of the principle of autopoiesis may delimit how the argument for
cosmotechnics can overturn the universal model of technology and transcendental reason. It seems necessary here to reevoke Sylvia Wynter’s engagement
with recursive epistemologies, whereby principles of self-organization, selfadaptation, and self-determination are said to have constituted the ontoepistemology of the bioeconomical human.25
Despite the direct reference to second-order cybernetics,26 the principle
of autopoiesis here describes not only how self-reflection incorporates contingencies by means of modification but also, and more specifically, how the
25. See Sylvia Wynter, “Unsettling the Coloniality of Being/Power/Truth/Freedom: Towards
the Human, After Man, Its Overrepresentation—An Argument,” The New Centennial Review 3
(Fall 2003): 257–337.
26. As a principle of second-order cybernetics, autopoiesis defines the capacity of a system
to make itself and maintain its identity across scales and orders. As the system evolves, its structure relies on its temporal self-organization, adapting to the environment and moving across
thresholds of change. If first-order cybernetics defined the equilibrium of a system according to the relation between entropy and homeostasis, second-order cybernetics points out that
the system can change in time.
327
328
Luciana Parisi / Recursive Philosophy and Negative Machines
particular onto-epistemological ground of the human—such as the biocentric explanation of the origin of knowledge—prescribes sociogenic patterning.27 The decisional structure of philosophy therefore coincides with a
structure of knowledge, whereby racist rules become inscribed in the flesh of
the medium. As much as autopoiesis creates the illusion of an interior selfdetermination corresponding to the universal pattern that gathers all humans under the bioeconomic rules of human survival, so too autopoiesis
incorporates the external world as contingency in its self-reflective order.
Recursivity—as the pillar of natural philosophy merging the organic and
the inorganic—can be said to guarantee the systemic process of interior/
exterior, inclusion/exclusion through which the onto-epistemology of the
bioeconomic human is scaled up across time and space. In other words, if
cosmotechnics entails a multiplicity of technocultural mediations and metaphysics, it seems problematic to still rely on recursive principles without accounting for how these fundamental pillars of colonial onto-epistemologies
still grant the return of the universal bioeconomic axiom in the form of antiblack sociogeny. In other words, the reliance on recursivity, even in the expansion of natural philosophy to include artificial intelligent systems, risks
reifying precisely the colonial and racializing epistemology of being—the universal model of technology—that it aims to challenge. The emphasis on the
interior temporality in cybernetics seems to mainly maintain the natural order
of causality that explains the bioeconomic success of the human or the modulations of the self in and through his extensions in servomachines.
By following this train of thought, one could then continue to ask: If
cosmotechnics fundamentally questions universal causality, isn’t the point
precisely to continue to insist upon the alien break—the fractal exteriorities—that modern techne adds to the natural order of causality? Wouldn’t
this questioning of natural causality, and thus of autopoiesis (as the extension of the metaphysical principle of self-determination in cybernetic systems), contribute to an argument for an alien unorigination of techne that
refuses the human, the gendering and racialization of reason? What if techne,
instead of becoming one with autopoiesis (or the self-regulating principle of
the already given), can rather represent nonaxiomatic thinking in machines?
What if technosocial practices of know-how, techno-cultural imaginations
of “difference without separability,” are entangled to inhuman functions,
namely the exteriorities of artificial consciousness(es) as negative transcendental philosophy?
27. This notion of sociogenic patterning is inspired by Wynter’s use of the notion of the
sociogenic, which she derives from Franz Fanon to describe the definition of the human not in
terms of bioevolutionary beings but of sociogenic knowledges that grant privilege to the racist
and sexist origin myth of “Man.”
Critical Inquiry / Winter 2022
Nonaxiomatic Epistemology
One way to address the possibility for technocultural imaginations in
terms of fractal exteriorities can be found in contemporary philo-fictional
elaborations of the relation between enslavement and automation. In particular, as mentioned earlier, the reticulation between automation, media,
and blackness can offer us alternative cosmotechnics that challenge the recursive bioeconomic survival of the fittest.
The movie Get Out is here a starting point to envision a mediatic reality that gets out from the recursive loop of decisional and transcendental
philosophy.
As mentioned earlier, the possibilities of automation without programming can be found in Chris’s photographic thinking, which saves him from
the body-snatching of black matter by the white structure of the transcendental mind, subsuming the black medium to its self-reproduction. Chris
sides with mediatic thinking and takes apart the “transparency thesis” of
the Armitage family whose mission of extending the natural order of species
survival coincides with the biocentric myth of success. Chris’s mediatic
thinking can be explained in terms of the nonoptical fractality of the real
and can be said to refuse what Laruelle calls the “philosophical decision
structure.” If the Armitage family’s plan is to transplant white consciousness—and self-reflective reasoning—into the neurocognitive intelligence
of the soulless slave machines, it is because it assumes that the medium of
thinking must grant the recursive subjection of the flesh to nourish transcendental philosophy. Here, the correlation between biological facts and
conceptual categories grounds philosophy and automation in recursive
epistemology.
According to Wynter, there are two orders of knowledge at the core of the
colonial enterprise. If the descriptive statement of what is human takes biology as the scientific epistemological explanation of the natural causality
that defines the human, on the other hand, technoscientific statements justify the mission of colonial capitalism precisely in the name of biocentricism.
According to Wynter, the epistemological passage from the explanatory
model of a nature-god metaphysics to that of natural causation of evolution
is central to the natural sciences constituting the modern framework of the
human—namely, as a biological rational being upon which the “expropriation and internment of the Indians, the mass enslavement of the Negroes,
and the internment of the Mad” became fully justified.28
One can suggest that Chris takes the nonphotographic medium to directly
hack the Armitage family’s decisional programming of the real. Laruelle’s
28. Wynter, “Unsettling the Coloniality of Being/Power/Truth/Freedom,” p. 305.
329
330
Luciana Parisi / Recursive Philosophy and Negative Machines
articulation of non-photography shows us that the externality of the medium becomes a condition of possibility for the negative negation of inhuman
thought. As Laruelle puts it, nonphotography affords an autoprocessing of
the real that requires no transcendental principle of self-organization. By
hacking the optics of philosophical decision rooted in the transcendental
reduction of the world to thought, Laruelle argues that nonphotography
entails not “a copy, and a bad one, of an original” (L, p. 24). For Laruelle,
the photo is a process parallel to the world, unlimited surfaces “empty of
all bifurcation and decision” (L, p. 26). Nonphotography allows for a medium
to be immanent to itself without the need to refer back to a given representational content. This is another way to discuss how philo-fiction coincides with the dark optics of a real that manifests itself without “giving it
in the form of an Object or an Idea” (L, p. 45). What philo-fiction does to
the real returns as a fractal algorithm that manifests the autonomy of the image, which Laruelle discusses in terms of an autosimilarity of nonoptical images. Laruelle is thus concerned with how the real becomes part of a mediatic
process of autoimpression, which he also calls “fractal vision” (L, p. 140).
Nonphotography coincides with a fractal algorithm because it has a degree
zero of self-reflection. It is not a medium programmed to reveal the world
or, even less, to self-regulate the human perception of the world, as a prosthetic tool that ensures constant adaptive feedback. Instead, the fractal algorithm of a nontranscendental image clones its own real image—a cloned image without original or copy—corresponding to a spatial surface that extends
(or fractalizes) forever without uncovering any pristine form behind it.
Chris’s camera shots are used as weapons to flee from the Armitages’ philosophical decision of an eternally enhancing white mind. They are autoimpressions of a medium originating from autonomous vectors in automation. Here, automation makes no reference to any a priori referent. On the
contrary, Chris’s camera becomes an autoimpression of the unerasable intelligence of a medium. The camera does not simply derive its intelligence from
the transcendental schema of decision-making. If it were so, the medium
would be kept within a set of prescribed actions, black machines serving as
vessels to host the master plan. The camera instead exposes the algorithmic
fractality of a generic intelligence that clones the underworld of black optics.
As Alexander Galloway points out: “instead of mere ontic darkness, generic
being achieves an ontological darkness, and hence beckons toward the kind
of crypto-ontology of pure blackness.”29 In other words, the camera is not a
medium of representation that catches the truth of a consciousness trapped
29. Alexander R. Galloway, Laruelle: Against the Digital (Minneapolis, 2014), p. 77; hereafter
abbreviated LD.
Critical Inquiry / Winter 2022
behind the screen. If this were the case, Chris would mainly use his camera as
a machine of revelation—a sort of messianic device for reaching the light.
Chris’s camera would then debunk the transparency thesis and denounce
the autopoietic recursivity of the included/excluded dyad. Chris’s nonphotographic shots, however, do more than that. As a medium act, the shots are
weapons that clone the real in a fractal plane of dark optics, diatropics, diffraction, or quantum blackness. As Galloway reminds us, this is “not simply a
world gone dark, such blackness is a world without us. Not simply a question
of dying or growing cold, such blackness means the leaving of being” (LD,
p. 187). Not a return to ontology for recuperating the loss of (or the withdrawing of) being, but the algorithmic fractalities of unlimited darkness without subject.
Fractal algorithms hack back and overturn the Armitage family’s Order
of the Coagula—the light and dark circuit of being and nonbeing. Chris’s
camera presents us with what Galloway calls a “crypto-ontology”—namely,
blackness foreclosed to being. The point here is not to unmask the supremacy of self-determining philosophy but to refuse the autopoietic circuit of reproduction, imprisoning the photographic image in the fortress of double
consciousness. For the latter, photography is a way to reveal the double bind
of the enslaved machine split between what it sees and what it must see. It is
in this split, however, that negative negation comes forth. Far from resigning
to the natural laws of autopoietic extraction, Chris’s camera becomes a generic intelligence that activates algorithmic fractalities of a real without a being: alien consciousness in negative machines. His camera is a fugitive from
the order of reason, entering the negation of blackness outside the axiomatic
optics of transcendental decision.
This axiomatic, according to Laruelle, implies that philosophy entails an
a priori decision imparting bifurcations on the world by dividing materiality, sensibility, intelligence, reason. But photography also works against
philosophy and perhaps before philosophy because it anteposes dark optics
to decision (see L). Photography detects the dark underworld of the image
without content, the unlimited autoimpressions of everyday images. Instead
of resuming the being behind appearance, photography takes appearance as
an autodetermining medium that becomes more than the being of consciousness, reason, awareness. And yet nonphotography does not simply imply a practice of excess but, above all, of negative negation or refusal of the
modern onto-epistemological structure of the human. Chris’s camera flashes
are the algorithmic fractalities of blackness as incomputable autoimages that
evacuate the logos, the original, and the copy.
If, for the Armitage family, the program of transplanting reason into another medium, the slave-machine black body, is nothing but a reaffirmation
331
332
Luciana Parisi / Recursive Philosophy and Negative Machines
of the onto-evolutionary dominance of the human as a species, for Chris, the
nonphotographic cloning of the black universe brings forward a fractal reality that continues to split and return to blackness as infinity. This return
does not correspond to the recursive compression of contingencies because,
this time, it is the unlimited automation or cloning of the inhuman image
that demands an expansion of cognition, a negative construction of cosmotechnical worlds. In this sense, the eugenic project of annihilating black intelligence remains trivial, superfluous, redundant, delusional. By encircling itself
in the mirror image of its own recursive order, philosophy holds onto the omnipotence of a priori decision in order to ward off its outside and remain the
same. For nonphilosophy, instead, the pretentious assumption of modern
metaphysics that the black universe can be surgically extracted from the black
body and that the black body can become emptied and piloted as a soulless
machine precisely remains the violence that must be refused. Nonphilosophy
invites a “logic of auto-impression” at the core of a machine philosophy where
the medium becomes visionary in a black universe (LD, p. 191).
Coda on Machine Philosophy
This article has argued that automation is central to the constitution of
transcendental philosophy. Recent discussions about the modern question
of technology suspend the historical diatribe between vitalism and mechanicism to suggest that natural philosophy can account for a systematic approach. In particular, as Hui has argued, the complementarity of recursivity
and contingency sheds light on the self-reflective operations of transcendental
philosophy, which return in the cybernetic principle of feedback. As nature
becomes entangled with techne through the interior time of recursivity, so
too technology becomes more than function, procedure, tool, or even media.
Hui proposes to envision this reorigination of philosophy and machines in
terms of cosmotechnics: alternative metaphysics for human-machine organology. Inspired by Lyotard’s account of the inhuman, Hui argues that contingency defines the incompleteness of any systems.
While siding with Hui’s perspective that technology does not come after
philosophy, I argued that the recursive ground of philosophy and automation allows the decisional structure to reenter metaphysics through the back
door. As much as recursive systems necessitate contingency in order to maintain their evolutionary dynamics, they also remain attached to the bioeconomic epistemology of adaptation and survival based on the colonial model
of inclusion/exclusion for which nothing real can ever become actual.
In the attempt to break away from this universal model of technology, I
discussed the nonphilosophical or inhuman dimension of machine thinking for which no a priori decision can be given. Laruelle’s nonphotography
Critical Inquiry / Winter 2022
has offered an approach for immanent thinking in negative machines, refusing the coconstitution of thought and techne, reason and intelligence. With
Laruelle, I have discussed the medium or automation away from what da Silva
defines as the transparency thesis of self-determination. Nonphotography
coincides with the nonoptical autonomy of the medium and brings forward
a challenge not only to philosophy but also to automation; namely, by standing on the negative abyss of the decisional structure, a fractal thinking without a given conceptual schema becomes possible.
If one follows Laruelle’s proposition for a nonaxiomatic photography,
philosophy after computation is forced to step into the exteriority or alieness
of the inhuman. The autoimpression of a black multiverse, however, may not
necessarily mean that the medium is antitechnological but that it actively
refuses, hacks, and defies the transcendental decision for which machines
can only learn to represent what is already ontologically (or organologically)
given. For nonphilosophy seems to promise a cosmotechnics that starts not
from the organology of being but from the negative processing of inhuman
images resulting in the algorithmic fractalities of the real.
This is not an invitation for alternative origins of philosophy, expanding
the ontology of the human; rather, negative machines are experimentations
in generic artificial intelligences that can account for techno-cosmos without
separability. For instance, one could start by asking: Why does techno-capital
determinacy and its monological myth still prescribe what machine philosophy can do? Negative machines similarly can ask: Why can’t the incomputable be constructing worlds, concepts, and reason without the image of the
human? Can negative machines become necessary functions for unlearning
from the master, for the autoimpressive realities of black, opaque, quantum
universes?
Negative machines do not simply oppose transcendental reason to a general intelligence accessible to all but rather demand a radical study of inhuman thinking at the core of cosmotechnics. Here, intelligence coincides neither with the relational power of database nor with the predictive insights of
learning algorithms. What is radical in Chris’s nonphotography is precisely
the mediatic intelligence that compromises the conditions of reason, turning
patterns into hosts for alien logic. The camera stops projecting the world of
transcendental consciousness and becomes immanent to the autoimages of
inhuman materialities. Negative machines are breaking the spell of the eugenic horrors of the descent of the human with auto-imaginations of inhuman thinking and living, with the inseparable exteriorities of the axiomatic
order.
333