**Links to**: [[Principle of Sufficient Reason]], [[The excessive, infinite, recalcitrant other]], [[The infinite other]], [[Learning]], [[Learning language]], [[Dialogical]], [[Intentionality]], [[Attention]], [[Sense]], [[Drive]], [[Desire]], [[Inclination]]; [[10 Bias, or Falling into Place]], [[Meat puppet]], [[Filter]], [[Aim-oriented]], [[Orientation]], [[Disorientation]], [[Vector]], [[Politics]], [[Class interest]], [[Self-interest]], [[Care]].
# The Principle of Sufficient Interest
 
### [[Postulate]]: The movements of [[Autosemeiosis]] are guided by the wonderful whims of interest, which are perfectly pedestrian and _degenerately_ communal (see also: [[Degeneracy]]).
 
 
>**Jeff Greene:** So, I’ve given up red meat.
>**Larry David:** Really?
>**Jeff:** Yup, no more red meat for me.
>**Larry:** Good for you. How come you’re doing that?
>**Jeff:** No reason.
>**Larry:** What do you mean “no reason”? You’ve gotta have a reason.
>**Jeff:** No, no reason. What do you care?
>**Larry:** Hey, schmuck-face, you can’t just say you’re giving up red meat, there has to be some motivation behind it.
>**Jeff:** No reason.
>
>Curb your Enthusiasm, Season 3, _Chet’s Shirt_, Scene 1.
 
 
>Thinking has interests that do not coincide with those of living; indeed, they can and have been pitted against the latter.
>
>Ray Brassier, _Nihil Unbound_, 2007, p. xi.^[Linking our interest in gravitational metaphors ([[10 Bias, or Falling into Place]]), the pull of the unknown ([[12 Negintelligibility]]), the strange concept of “seeming” and learning one’s interests (this chapter); also Brassier: “My interest gravitated early on towards ... figures like Hegel, Heidegger, and Sartre. I did not understand them, but their difficulty seemed glamorous.” Ray Brassier, interview with Richard Marshall, https://www.3-16am.co.uk/articles/nihil-unbound, published Feb 2021, accessed Feb 23 2025.]
 
 
<div class="page-break" style="page-break-before: always;"></div>
# The Principle of Sufficient Interest
 
**Summary: reasons and interests.** The pervasiveness of the concept-phenomenon of _interest_ seeps into every aspect of sociocultural life: “_this_ is more interesting than _that_,” class interests, conflicts of interest, dilemmas of self-interest, etc. The whims of what we call ‘interest’ track our inclinations. The specialized, theoretical concept of _interest_ goes by many names, in reference to all kinds of intentionalities (e.g., motivation, drive, concern, in, e.g., respectively: relevance, psychology, care ethics). In what follows we will tie colloquial and theoretical approaches to interest into a few arguments which reveal interest as a fundamental and unavoidable factor in the process of language-learning, and thus also of what we call _reasoning_. Interestingly, as a notion denoting the contingent specificity of a (person’s) particular focus or inherent subjective disposition, interest is often—though not always explicitly—contrasted with what ought to be considered the transcendentally detached or disinterested character of _reason_. Though the many possible interests of reason—and _interested reason_ as a concept in itself—have been topics philosophy has treated at large since Kant, most modern accounts of reason or rationality—particularly in AI, and especially by way of the concept of _bias_ as something to be avoided, or by way of proposals of “generality”—have attempted to unify its procedural, formal, inferential, logical, etc., aspects into reproducible models, in order to _ground_ accounts of (scientific) objectivity which promote different degrees of realism which ought to be _disinterested_. Historically, reason’s most well-known self-immersive alienating attempt at this is no other than the _principle of sufficient reason_, to which this piece owes its title. The modest and boring intention of this piece is to insist that, however aspiring toward abstractive detachment, reason, as an iterative and procedural (i.e., computational) social activity, is inevitably tied to a process of _interested learning_. This process, owing so many of its (practical) determinations to the vastly divergent and situated paths of those who perform-endure it (adding up to “cumulative cultural learning”, Tomasello 2016), can never be said to converge towards the realization of a common (transcendental) reasoning goal. Except, perhaps, when the goal is *learning* (i.e., _function-transfer_) _itself_, and when we witness language as supervenient over dreams of voluntarist individual sovereignty over it. This is our main claim.
<small>Keywords: Interest, inference, reason, dialogical reason, PSR, learning.</small>
<div class="page-break" style="page-break-before: always;"></div>
 
### Introduction: Interest and its reasons for being here, why should one _care_?
 
>The etymology of the term _inter-esse_, “to be between,” [...] marks the annihilation of the distance between the person and the materials and results of [their] action; it is the sign of their organic union.
>
>Dewey, _Interest and Effort_, 1913, p. 17.
 
>The problem resides in the fact that the faculty of thought is identified with “the system of judgment,” and **knowledge with the model of the proposition**. Whether from its phenemenologico-constructivist or cognitivo-instructionist wings, contemporary anthropology has long discoursed on the severe limitations of this model in accounting for intellectual economies of the non-Occidental variety (or, if you prefer, of the nonmodern, nonliterate, nondoctrinal, and other “constitutive” absence varieties). In other words, anthropological discourse has devoted itself to the paradoxical enterprise of heaping proposition upon proposition on the subject of the nonpropositional essence of the discourse of the others, **going on endlessly about what supposedly goes without saying**.
>
>Viveiros de Castro, _Cannibal Metaphysics_ (2009) 2014, p. 79, our emphasis in bold.
 
The development of something supposedly deserving of the name of “artificial (general) intelligence” bears witness to the difficulties presented by a highly interested reason. Dreaming the image of a disinterested reason, tasks and mileposts such as the development of “explainability,” “(perspectival) frame,” “common sense,” or even *apparently straightforward* natural language understanding, pave the way with a variety of interested obstacles that were rationally unforeseen, due to the pursuit of one-dimensional, unconditional or universalist solutions (see, e.g., Huang and Liem 2022, Crawford 2022, Bender and Gebru 2021, Mitchell 2019). In short, try as we may, reason will always be perspectival.^[See: [[Vantagepointillism]].] What will be hereby proposed is that by paying closer attention to the concept of _interest_ as it occurs dialogically and, more specifically, pedagogically—interlacing complex processes in the biology of mimicry, the psychology of imitation, historicomaterial contexts, the sociodynamics of relevance, dialogical alienation, Luhmannian double-contingency, symmetries, and more—a new tracing of the processual nature of _reason_ can emerge, one which can inform a different concept of “intelligence” in the pursuit of AI. One which, as will be shown, gives prominence to the idea that learning is primarily interest-encoding, an argument which reinforces existing proposals of the socially-distributed nature of dialogical reason (Dutilh Novaes 2020, Vasil et al., 2020), its always-incomplete and unstable character (Whitehead, Feyerabend), and its inevitably aesthetic (and cruel) dimensions (Nietzsche).^[See also: [[Aesthesis]] and [[Digestion]].] Several authors are presented for the sake of frames, examples and elucidations, in particular Wilfrid Sellars’ account of _language games_ is brought forth, as it ties dialogical learning, reasoning and interest together into one argument (which we partly uphold but also contrivedly deconstruct). Developmental psychologist Michael Tomasello, in an article concluding many of his thoughts on evolutionary cultural transfers, ends on the note: “The interaction between cultural learning and learning culture is a key question for future research.” (2016, p. 650). We claim that this interaction is the transfer of **pure function**,^[“Thus the metaphysical exploration of the structure of being can only be carried out in tandem with an epistemological investigation into the nature of conception. For we cannot understand what is real unless we understand what ‘what’ means, and we cannot understand what ‘what’ means without understanding what ‘means’ is, but we cannot hope to understand what ‘means’ is without understanding what ‘is’ means.” (Brassier 2011, p. 47).] which the concept of interest is able to track, if we follow its linguistic motions.^[This observation takes from ideas in systems biology and evolutionary theory, and explores _language_ as a mutating, evolving system. Mostly, similar to the arguments made in [[09 C is for Communism, and Constraint]], this piece aims to **counter** the disappointment in the image of the individual organism as one _losing_ “agency” in the face of the systems it exists embedded within. The voluntarism expressed in this disappointment leads, we think, to oppression. Without it, or with less of it, social creatures can attune to each other more exploratively. This disappointment is found across the board, from Aristotle to Schrödinger, for example, presented by Felin and Kauffman as problems: “biologist Patrick Bateson notes that “the picture of the external hand of natural selection doing all the work is so compelling that it is easy to regard organisms as if they were entirely passive in the evolutionary process” (2017: 105; also see Noble and Noble, 2017). In his book _What is Life?_ Erwin Schrödinger points out that “popular expositions of Darwin’s theory are apt to lead to a gloomy and discouraging view on account of the apparent passivity of the organism in the process of evolution” (1944: 106). These problems have also been echoed by others more recently (e.g., Niemann, 2014; Longo et al., 2016; Montévil, 2018; Noble and Noble, 2018, Odling-Smee et al., 2003; Peck, 2019; Vladar et al., 2017).” (2019, p. 4). A viable road towards truly _social_ coexistence is one where the system’s function-evolution is acknowledged _over_ that of the individual, we think.]
Interest might be the most dialogically-relevant of volitions. In learning about each other through dialogue, people tend to frame their mutual degrees of interests in things and, more often than not, this is with the intention of arriving at the possibility of overlapping perspectives. Once interests—and therefore, sometimes, intentions—are revealed, or somehow demonstrated through a slowly unfolding alienation; a relationship of double-contingency,^[It is the symmetrical or “like-minded” (our terms) fallacy which Luhmann deconstructs, consensus is neither guaranteed, nor normatively constrained (see: Vanderstraeten 2002). Luhmann: “Beginning is easy. Strangers begin by reciprocally signaling each other indications of the most important behavioral foundations: the definition of the situation, social status, intentions. This initiates a system history that includes as well as reconstructs the problem of contingency. As a result, the system increasingly is occupied with arguments about a self-created reality: with handling facts and expectations that the system itself has helped to create.” (Luhmann 1995, pp. 131–2). For Luhmann, double contingency is what enables social interest and communicative creativity, it is not a problem to be solved by cultural consensus, as Parsons noted, and as much of the contemporary literature on social interaction continues to rest (see [[Sameness]] and [[The infinite other]].)] a lot of what we consider _sociality_ can be said to start. In contrast with (traditional theoretical) reason, which ought to have a communicable underlying logic which “applies to all”, someone’s _interest_ is singular, specific, and can be lifelong-settled or possibly ever-changing. However, it is in finding interest at the bottom of the processes of dialogical reason, that we will thus give it explanatory primacy, as a formulation which schematizes^[_Schematizes_ in the sense that it converges these into a generality. See [[Schema]] for a more detailed account of this.] sense-making, attention, drive and desire into a _meaningful yet relaxed_ argumentative premise. _Relaxed_ because interests can change through time, there is no ‘formal’ requirement that a particularly interested perspective remains static. At the same time, it is _meaningful_ because of this very same reason: it is _pragmatically_ truthful.^[The pragmatic definition of [[Truth]] famously being: that which a community of inquirers agrees upon, for the moment being. A kind of consensus but based on the idea that it is always subject to change, anarchic (in the Whitehedian sense).]
An excellent demonstration of the pragmatic trajectories interest _as_ reason may travel, is presented in Lorraine Daston and Peter Galison’s _Objectivity_ (2007), where the authors trace how scientific representation and ideas of objectivity change as different paradigms of interest co-evolve with historical and technological developments. Something Peter Dear notes in reviewing the book is that it does not go far enough in treating the issues of pedagogy and training in the making of objectivity paradigms (Dear 2012, p. 14). He also notes that “Scientists, both _avant_ and _après la lettre_, appear in this book as creatures with rich inner lives, but with little social integration of any kind.” (ibid.). It is the anthropophilosophical intention of this piece to focus on various aspects of how this type of interest of reason (to arrive at, e.g., scientific consensus) is socially integrated, encoded, by attending to its pedagogical lenses, which we treat in the later sections of the chapter. We first cover the concept theoretically, then we relate it to the development of AI, then to dialogical unfolding through Sellars, in order to arrive at pedagogy, and the conclusion.
Interests can be made rather explicit.^[See [[10 Bias, or Falling into Place]] for an analysis of _whims_. In researching the etymology of _whim_, one will discover its interrelation with the concepts of [[Fantasy]], [[Inclination]], [[Determinacy]], and thus [[Gravity]] (i.e. the ultimate [[Ground]]).] What interest is also able to reveal is our naturally _inclined_ condition as, among other things, gravitationally-bound and distinctively chemically-predisposed organisms. But, of course, it does not have a _center_, as it unfolds doubly-contingently, bidirectionally, with its self-interested niche: it emerges from and results in the (meta)linguistic, sociopolitical tensions _exhibited-promoted-performed_ by the given historical timeframe, current spectacle or _hegemonic chunk_. Reza Negarestani frames the Hegelian unfolding of reason through a kind of disinterest, as the autonomous capacity of critical self-consciousness to dialectically become _other_ than itself, by recognizing what it is not (2018, p. 274). We take a more traveled, pedestrian road here and simply attempt to show ways in which the colloquial concept of interest can help explore how (trans)individuation moves through various fitness landscapes of learning. Owing to its unavoidably situated character—how it _situates_ characters—it can be made rather explicit (i.e., interests can be pointed at). This might help promote sociorational^[We use “sociorational” to promote the social image of reason as opposed to its singular-agent character, which seems to be the one most often promoted, especially in pedagogical settings, where individuals are assessed, well: individually.] awareness even more so than it already does in its colloquial employment. In general, as interest ties attention with decision-making and action, it is highly revealing of the mechanics of active inference that this speculative project is largely invested—and interested—in.^[More on this in: [[E Pointing]].] It is in hope of elucidating biases, proclivities, historicomaterial contexts, etc., that this piece turns to the concept of _interest_ as that which can reveal what is usually plastered over with an air of objectivity and transcendentality in the concept of _reason_. This may already be obvious to the reader. What we will entertain, as a _divertiment_, is a thought experiment in the sense Viveiros de Castro presents it here:
 
>... not in the usual sense of thought (imaginarily) entering experience but, rather, of the entry into thought of (real) experience. Not the imagining of an experiment, but an experimentation with the imagination or an “experimentation with thought itself.”^[Viveiros de Castro cites François Jullien here, Jullien and Marchaisse 2000, p. 71.]
>
>Viveiros de Castro, (2009) 2014, p. 187.
 
We will try to change the concept by overloading it.^[Our observation is not dissimilar from Felin and Kauffman here, on the “environment”: “The generic word “environment” masks ... specificity and ... munificence. Environment has become a catch-all term for explaining behavior and survival. But the idea of an environment needs to be unpacked. Environments are impossible to fully represent, though the sciences often (implicitly) assume that this can be done, in the form of assumptions that are made about omniscience and an all-seeing eye” (2019, p. 7). We treat the concept of _interest_ in the same way, in order to highlight aspects of its situated specificity, and the generic search for function it reveals.] Every time the concept appears on the page, try to become aware of what it is pointing at, what it is doing-predicting-producing in the context, how it speaks to you, or does not. We will treat the concept of interest through an anthropophagic lens,^[A radical anti-anthropology which sees itself as consuming, incorporating, regurgitating and being subject to the same massacre.] as a philosophy which betrays itself,^[Marder: betrayal has a paradoxical structure: “with its interwoven strands of a surplus disclosure and a breach of trust.” (2020, p. 79). In this text Marder also proposes an experiment not unlike the one proposed by Viveiros de Castro: “The experiment I am proposing on these pages is not self-contained: once the flood-gates open, the work of thinking and saying, phenomenological acts and the very notion of truth, will change beyond recognition.” (ibid., p. 80). We commit to a slightly more modest starting point, that of modulating the concept of interest.] and witness the interest concept-phenomenon as something which learns about itself, through this text, rendering the author a kind of conjurer. It is interesting to pay attention to the concept in everyday conversation, as well as in academic argumentation. In both cases, ‘interest’ comes up whenever there is nothing else to say about the reason(s) for something being the case, it becomes a principle of sufficient reason. It is often presented as a basic way in which action-oriented agents organize themselves and function towards structures of information, aesthetic affairs (“I’m interested in this but not in that”) or, similarly, in how structures or affairs may be revealing of something yet to be determined or explicitly formulated (e.g., discussion sections in scientific papers are littered with “interesting” future inquiries, despite the Mertonian norm that disinterestedness should be central to scientific inquiry as “‘[it]’ has a firm basis in the public and testable character of science” (Merton 1973, p. 276 quoted in Rieder 2020, p. 254). A different but related proposal is the Popperian aim-oriented empiricism of Nicholas Maxwell, which advocates for confrontation and explicitation of the metaphysical interests underlying scientific methods and paradigms (Maxwell 2005).^[We cite this 2005 work but there exist earlier proposals by Maxwell, as well as more recent ones, too.] This approach also shares aspects with Guest’s metatheoretical calculus (2024, treated in the 001 Introduction to the Poltergeist), another proposal suggesting that we make interests explicit.
 
>We are betrayed by language, which empowers us to express ourselves and frustrates the very expression it warrants. ... No one appreciated better than Hegel how the mutually exclusive significations of a word—e.g., _to bolt_, _to ravel_, _to sanction_, _to cleave_, and, naturally, _to betray_—that are in the same breath said and unsaid (contra-dicted) itself reflected the dialectical essence of language.
>
>Marder 2020, p. 85.
 
Or:
 
>To insist that estrangement has already taken place is to realize that the recurrence of originary dispossession is what enables us to take possession of ourselves and to affirm the necessity of this possession _knowing that it entails further dispossession_. History dispossesses us even as it provides us with the sole resource for becoming free.
>
>Brassier 2019, p. 104, (our emphasis in italics).
 
Another question our proposal is thus immediately presented with is whether interest is to be deflated and taken for granted in everything and anything we talk about—its banal colloquial omnipresence already reveals this platitude—or whether to further estrang(l)e it as an interesting vector of specialized inquiry. Further questions we could gravitate around are: are we always _inherently_ interested^[See, e.g.: [[Depression]].] in order to self-evidence and survive? In which situations do we laboriously “cultivate” and “refine” our interests? Do we explicitly present each other with evidence of things which interest us and things which do not, in order to test, align, challenge or reinforce our collective intentionality? When and how do interest-revelations signify predictive social coherence (Vasil et al., 2020) and when and how are they, in fact, the testing ground of social indeterminacy (Luhmann (1995 (1984))? And, generally, how can the notion of _interest_ help us think about how we think, specifically when considering the transfer of _reason_, between beings, as structured by highly-interested linguistic encodings? While the former certainly guide our inquiry, we will mostly focus on the last question. What is obvious, and therefore annoying, is that the concept is omnipresent, and very vague, serving multiple sociorational predictive functions simultaneously—it can denote immediate or longer-term attention; motivation; curiosity; investment; benefit; and bias, among others. This polysemy will allow us to bridge phenomenological experience with sociopolitical structure.
The reasons for presenting interest as an interesting explanatory phenomenon are various: first of all, it is in the interest of this project that we observe language-learning differently in order to become more aware of our collective unfolding within it, and its possible modulation. Proposing ‘vectorial’ updates of the meanings of word—to transpose a metaphor from NLP—is not only an always-already-occurring activity but an absolute necessity for the meta-evolution^[Evolution recursively unfolding within other evolutions: material, biological, technological.] of concepts. The remarkable thing about the concept of _interest_ is that it is **itself** a vector of attentional alignment, it highlights perspectival attention, in language it tells attending agents what the (distant) predictive directionality of a claim is: ‘more or less in that direction.’ The necessary invention of new concepts begins with the examination of existing ones, and with a capacity to understand language as subservient, first of all, to itself,^[And it shares this with (its) music, see [[Modulations]], [[Language-modulating]] and [[Music as permanent revolution]].] and not always limited to teleological, utilitarian senses, or as a mere representational medium always serving the pursuit of other activities. Secondly, in contemporary discussions, unlike concepts such as drive, bias or desire, _interest_ seems^[More on this elusive, probabilistic verb later.] to be more “_general_”, everyday-colloquial, perhaps less loaded with past analyses, and thus more ripe for our modulating.^[See also: [[On the importance of vegetables and sand for philosophy]] for an elucidation of this vegetable metaphor.]
Interest—not unlike desire but also _very_ much unlike desire^[Again, because interest can be made quite concrete, whereas desire is infamously bottomless.]—is also often framed around _minor_ concerns (e.g.: your interests may differ from your profession), or applicable in situations where bias and subjectivity play an obvious role in the determination of decisions (e.g.: a company’s interests are driven by profit).^[Coincidentally alluding to _interest rates_ as well, which is not our main interest here.] Of course, in relation to desire, interest also pertains to much of what has been theorized as _will_, _volition_, and (congregational) _affect._^[Following the famous line by Spinoza, presented by Deleuze and Guattari: “We know nothing about a body until we know what it can do, in other words, what its affects are, how they can or cannot enter into composition with other affects, with the affects of another body, either to destroy that body or to be destroyed by it, either to exchange actions and passions with it or to join with it in composing a more powerful body.” p. 257, _A Thousand Plateaus_ 1987 (1980).] But while we will sometimes touch upon these concepts, they will not be our focus. They have been, diametrically as well as dialectically, confronted with reason in a lot of theoretical investigation—which can distract from our analysis of interest as dialogically reasonable, or reason as interestingly dialogical.
“Interestingly, recent research has also found that young children themselves engage in instruction at a much younger age than previously thought.” (Tomasello 2016, p. 647) This attention-function—“vectorial” or _indexical_, in the sense of pointing—of interest manifests through what Tomasello calls “transmission chains” (what we later call _interest-encoding sequences_), where children learn and subsequently teach others by pointing out to others what they have been shown, often ignoring possible ways to do things, they seem to “stick to the program” much more than exploratively search for novelty. We will see how in our images of playing chess and actively engaging in reasoning, we often forget how encultured we are, and how interests have been encoded _onto_ us, where the functioning of a “parsable” and coherent language that is socially-acceptable often takes precedence over any experimentation or exploration. In studies of norm-enforcement (Tomasello cites e.g., Rakoczy et al., 2008), cases where, e.g., 3-year-old children corrected norm violators “(e.g., “No, it doesn’t work like that”) ... instruction quite often was formulated in generic language as well, for example, “These things go there!”” (Tomasello 2016, p. 647). These indexical ways that young children use language for instructing others pedagogically reveals how coercion and conformation are rather prominent, in function-transfer.
The pedagogical authority children seem to wield, already at a young age, demonstrates, for our argument, how interest, as that which modulates attentional mechanisms, can help expose the interests of language itself in passing _as function_ from one body to another. We cannot say the children have any real interest in rule-enforcing upon others, it clearly seems to be a case of rule-enforcing itself, as a function, transferring across beings. When children receive instruction they more often than not seem to **adhere exclusively to what is prescribed usage**, ignoring other _potentially interesting_ applications (Tomasello 2016, Bonawitz et al., 2011): “Indeed, this authority is so strong that if a child is instructed in how to use a novel artifact (which has several possible functions), they tend to stick with that function almost exclusively and so ignore its other interesting functions.” (Tomasello 2016, p. 647). Interest-encoding thus emerges as the functional infrastructure of cultural transmission—the process by which language, as a complex network of indices, programs, the very attentional, perspectival vectors through which future generations come to navigate conceptual-material landscapes, revealing the transfer of **function** (i.e., of learning itself, what Tomasello calls a “cultural ratchet”, 2016, p. 647), more than anything else. “Pedagogical learning [i]s cultural learning in the sense that its authority emanates from the cultural group and its institutions, as larger and objective realities that predated the child’s arrival on the scene.” (ibid.).
 
### Pure interest in itself as the ground of reason
 
>Whenever we find ourselves among double binds, our aim should not be to get to the truth as quickly as possible but to invent a new language ... that maximizes friction in such a way that the various parties at war slow down and effective aggregation in affective learning processes becomes necessary.
>
>S. van Tuinen 2019, p. 102.
 
Generalizing from complexity and unpredictability towards grounded principles is an effort in the interest of habituation, or prediction, even when this generalistic reasoning is aware of the decision that the cut made in said things (e.g., in complexity) is but a mere _placeholder_ for the sake of persistent self-evidencing.^[See: [[12 Negintelligibility]] and [[X Simplifying complexity and complexifying simplicity]] for an exposition of this.] Generalizing collectively, through language, means some agents will be learning, others will be demonstrating the logic and application of principles, and the vast majority will be sitting somewhere along the gradient.^[E.g., in musical experiences, for example: where improvisation leads to distributed situations where it is not clear who is learning or leading.] This is why, the way something as fundamental as a _principle of sufficient reason_ (PSR) is dialogically articulated, is crucial for a sense of what _interested reason_ precisely entails. How does a contingently unfolding agent absorb the predictive traction of, i.e., the cognitive alignment with, something as abstract as the PSR? The fact that children never stop asking “why?” should give us a hint.^[Side-note: in Spanish “why” is “por qué”, which we can translate as _from_ or _for_ or _through_ what. To me, this rings more PSR-oriented than _why_, which feels more open-ended. Same in Dutch with “waar-_om_” (where-_from_ or _around_, perhaps closer to _whence_ in English). The etymology of _why_ relates to _what_ and _where_, too, hinting at our language orientations mapping onto about our spatial-reasoning.]
Much of what has been prescribed as _reason_ has tended towards the disinterested and ‘transparent;’ the transcendental. In the paper “Kant, Fichte and “the Interests of Reason””, Breazeale states how _interests_ tend to signal the presence of **conditioned** needs, whereas reason is supposed to be unconditioned and disinterested (1994, p. 81). Breazeale identifies Kant’s interest in the elimination of (self-)interest from the functionings of reason (KRV, A475/B503), and also observes that Fichte’s reaction to this was that without interests, an inevitably conditioned and _finite_ being can find no rational orientation (either in theoretical or practical reason). For Kant, in the case of empirical interests: something external motivates their motions, whereas the _pure_ interests of reason “spring from and express the nature of reason itself.” (Breazeale, p. 83). Thus, in the interest of disinterest, Kant’s reason, Breazeale observes, strives towards its own unity, which cannot be derived from anything higher and can thus be said to strive towards the unconditioned, the disinterested, in pursuit of the ultimate objects of reason: “the immortality of the soul, the freedom of the will, and the existence of God” (ibid. p. 84). This pursuit, impossible when faced by finite agents and thus famously leading us to antinomies, can nevertheless be understood as having a _regulative_ role in finite practical endeavors. (Part of) the interest to act in accordance with a universal moral law, is derived from the law having been observed purely rationally, _as_ law, as ‘universally applicable.’ In our context (and lining up active inference with Kant, see also: Swanson 2016), this universality implies predictive traction.
However, in our argument, the focus is on the impossibility to _suspend_ interests, and to understand how the concept of interest itself determines learning and _reasonably disciplined_ behavior. Not only because the way laws are _learned_ is by testing their outer edges—exploring their boundaries and thus occasionally breaking them—but because they are passed on from human being to human being by a method which rests primarily on immediately practical and conditioned interests, and prior to any conceptual inclination towards notions of infinity or universality. Rules of possible conduct, such as the grammar of (language) games, are relevant in the attainment of social and thus pedagogical ends, if observed _as_ rules learned through the constraints of specific interests, as will be seen.
 
>The generation of rules and the capacity to reason are inconceivable without interaction, and are inseparable from the complex contexts that arise throughout its course. However, contra Sellars, this interaction is not a matter of acquaintance with norms as a matter of cultural evolution; and contra Brandom, it is not an index of a substantive sociality of reason. It is rather the very formal condition of language and meaning – a logico-computational dynamics that realizes the syntax-semantics interface. The substantive sociality of reason is built upon this formal condition, not the other way around.
>
>R. Negarestani 2018, p. 376.
 
What this implies is not only that learning is a situated and highly conditioned affair (despite the institutional insistence that all students be judged by a flat standard), but that in the _learning_ that is implied in the use of any language: every single word a speaker employs (every single ‘move’ that is made in the game) is directly revealing of a learned interest. This simple observation responds to the pervasiveness of concepts such as “competence,” in pedagogical settings, which in many cases is seen to imply an end to learning, and responds to the apparent social agreement that words are mere _vehicles_ which serve higher-order activities.^[This last point is most explicitly made in Deleuze and Guattari’s criticisms of Chomsky.]
In our argumentative diversion that reason and interest are not indifferent to each other, that they share a _grounding-structuring-underpinning_, how they complement each other’s movements becomes through learning becomes our point of attention. We can trace this complementation, too, to the _acting_ or _willing_ part of Schopenhauer’s PSR fourfold: the motivational, the willing drive to _undergo_, _observe_ and _know_ in the other three aspects of his PSR (respectively: becoming, being and knowing). The conceptual-linguistic motion of interest and reason is bidirectional (as far as a non-linear but retroactive conceptualization allows): at times reasons find their interests, and at times interests find their reasons. When a piece of text is read without motivational interest, it reads, but it might read like a string of _word-sounds_. When one reads it interestedly, it opens up a grounded vantage point, as it relates those _words->sounds->concepts_ to interesting reasoning possibilities. Massumi reminds of Hume on how reason can ground plenty, but not _its own rationale_: “it can give no ultimate reason why reason is better.” The regulative role of reason, as social discipline, is to temper collective _chaosmos._^[To cite _Finnegans wake_, Joyce 1939, echoed later in Guattari.] But why?
The rational quest to overpower the affective shows, in effect, how reason itself is a (stubborn) passion. On reason as self-propelling instability, Massumi notes that Hume also reflected on how nothing in reason’s own operations can ground it in any certainty as to its limitations (this is the awakening of Kant): much like the inability to determine the halting of a simple program on a Turing machine.^[More in [[Intuitionism]] and [[Choice sequences]].] “Without the corrective of doubt, affectively propelled action will often go awry. But without belief in the world, action is impossible. We are always plunging headlong into events” (Massumi 2019, p. 4). In an act of interested reading, any concept is reframeworked, reformulated and reduced or aggrandized beyond what an iterator could have intended.^[E.g., I mentioned “diversion” earlier, on the surface use it is a distraction strategy, and/or a transductive move which splits open a new (interpretative) path: creating diversity. But the meaning is also that of _something that entertains_. How concepts learn, how etymological fitness landscapes reveal interested learning paths, is evidence to why we should break and reconstruct concepts.] Text; wording, like other things interested reason invests itself upon, happens to become something different than what it is, by the fact that it is read. These possibilities are not ‘clear’ protruding affordances that _any_ blank reader would notice, but are, time and again, subject to highly specific contextual variations, or enabling constraints, which promote learning and novel thought. At the same time, a preliminary word of caution: this is not meant to depoliticize interest and promote it to a banal and relativist liberalism. We will move through different thoughts where reasons and interests coincide, in order to observe that _learning_, a core drive in the interest of reason, necessarily calls for the dialogical; social aspects of reason’s interests to be highlighted, a ground so fundamental that it often goes by unmentioned. A territory that needs to be highlighted as it incessantly eroded by many of our methods and technologies: from the individualistic educational frameworks we uphold, all the way to the narcissistic domes of “social” media.
One of the things that tends to make philosophy _more interesting_ is when we find out about the interests of the lives producing it. That is, when we understand what a philosopher _wants_ from life, what their supposedly non-philosophical interests are. Entire _other_ lives are spent figuring this out about authors passed: hence the biography of Kant tells us something interesting about his philosophy. In language use, this recursivity—i.e., revisiting a function, seeing the life _as a life_, re-entry into the message _as_ message—reveals itself in how we entertain _metalinguistic_ functions: not just to talk about the fact that “I am having a thought” but about what a thought and _having_ it _is_. “What, stated in terms of the model, is the scientist doing?’ The first, and most general, answer to the question obtrudes itself: scientists are individuals who are creating _new language points_. ... real science consists [partly] in creating new analogies, and in making lucky and unlucky guesses.” (Masterman 2005, pp. 79-80). Understanding this makes conditions _apparent_, and therefore malleable. Social discipline schemas—science, the court of law, the university, the family—can sometimes, therefore, provide evidence of interests: their persistence reveals their constraints, enabling meta-reflection by grounding us in rules.^[More on language and rules through Sellars, later.] But more often than not they perpetuate the postponement of meta-engagement without in fact admitting any self-consciousness with regard to its postponement, in other words, they play by “business as usual”.
This sociality of thought creates collective vantage points, but often “we” turns out to be something more contrived and hidden than “we” commonly assume in everyday speech.^[See: [[Who is we and what is why]] interview with Gerard de Zeeuw. Tomasello’s work also repeatedly points to how collaboration is the standard, not the exception, which makes us question standardized, individual testing even more.] Pronouns such as _we_ serve as a functional marker of (self-)interest, so to question _which selves_ pronouns apply to in the context of learning and interest is a reasonable way to begin, as we will see below.^[And as is explored in [[Modulations]].] Philosophy as the love of knowledge is painted in its most “virtuous” light, after all, as the love of self-knowledge _for the sake of_ a speculative and collective coherence: of learning together through the linguistic proposals of a given chunk of time.^[See related ideas in: [[Invention]], [[Discovery]], [[Noise]].] As van Tuinen (2019) argues, the PSR is significant in a recursive, unstable, ecological reason, where its ground is necessarily always external, and therefore alienating _and_ fueling. This is what van Tuinen, following Stengers, following Deleuze, calls ‘thinking by the milieu’. It doesn’t always work out, sometimes the milieu _thinks_ one:
 
>The imagined “they” constitute a kind of invisible conspiracy of masculinist scientists and philosophers replete with grants and laboratories. The imagined “we” are the embodied others, who are not allowed not to have a body, a finite point of view, and so an inevitably disqualifying and polluting bias in any discussion of consequence outside our own little circles, where a “mass”-subscription journal might reach a few thousand readers composed mostly of science haters. At least, I confess to these paranoid fantasies and academic resentments lurking underneath some convoluted reflections in print under my name in the feminist literature in the history and philosophy of science. We, the feminists in the debates about science and technology, are Reagan era’s “special-interest groups” in the rarified realm epistemology, where traditionally what can count as knowledge is policed by philosophers codifying cognitive canon law.
>
>Haraway, 1988, p. 575.
 
How does our cognitive canon ultimately ground itself? If we orient ourselves by way of patterns such as gravity,^[As treated in [[10 Bias, or Falling into Place]].] then what of other degrees of spatiotemporal orientation? A pattern most basic is our given physiological symmetry (see: [[All things mirrored]]), and the fact that we can turn left or right (think of this as abstractly as possible, see also: [[Line]]). Change, herein, is our condition ([[Choice]] herein is change). Orientation is the _embeddedness_ within this unavoidable change. Try as we may, it is very difficult to imagine oneself as lacking this very basic sense of orientation.^[Though some altered states elicit this.] Kant asks us how we could possibly “procure sense and significance for our concepts”^[“_Was heisst: Sich im Denken orientiren?_”, October 1786, _Berlinische Monatrschrift_ VIII.] without an intuition, which, “ultimately must always be an example from some possible experience” (ibid.). _Some possible experience_, is what we are after. The fact that *it* does not have an example, and that _it_ remains potentially impossible, in its speculative balancing act of _possibility_, tells us precisely what we need to know: some concepts _seem_ conceptually precise because they are not appended to representations of possible experiences. Unbalanced imprecision is fruitful.
### Learning and style
In the context of learning, John Dewey already made the crucial remark in 1913 that learning as guided by _interest_ is functionally entirely different from supposedly disinterested learning. Dewey argued that “It is absurd to suppose that a child gets more intellectual or mental discipline when he goes at a matter unwillingly than when he goes at it out of the fullness of his heart … It is psychologically impossible to call forth any activity without some interest.” (pp. 1-2, Dewey, _Interest and Effort in Education_, 1913). For Dewey, interest as intentional, “undivided attention,” is absolutely necessary for the satisfactions of, and perseverance in, learning.^[Dewey responds to a tradition of devoted _effort_, which he opposes to (creative) interest, and while this certainly resonates with disciplinary traditions of past centuries, this cult of disinterested devotion is in fact still very much with us in the guises of meritocratic self-application and uninhibited (professional) competition. Contemporarily we could also characterize/diagnose as the obsession with “resilience”.] As Dewey points out, learning—and, we will argue: (meta)language-learning—involves interest in the form of desiring-drive as a crucial component in developmental nourishing. In a similar vein, relating style to drives and desire, interest also finds its place in Nietzsche, by the hand of Sarah Kofman in _Nietzsche and Metaphor_, ((1972) 1993, pp. 2-3):
 
>All style reiterates a primary writing, that of the ‘drives’. Thus it is as vain to seek to impose a canonical model on writing as it is futile to seek to legislate universally in morality: **each _must_ do only what he _can_**.^[Our emphasis in bold. Translator notes he chose “drives” for the French “instincts”, in light of Freudian _Triebe_. My emphasis in bold. Paul Patton also cites Miller on Kofman on Nietzsche: “she was at the forefront of those interpretations that led to a renewed appreciation of Nietzsche as the creator of a style that was "neither purely conceptual nor purely metaphorical, neither scientific nor artistic" (Miller 1985, 47)”. Patton 1999, p. 97. We hope to be tuning to this, too.]
 
This relates what will be briefly explored later when reflecting on Kant and (the learning of) universality, as well as to what Dewey frames as “psychologically impossible.”^[Related: “The genuine principle of interest is the principle of the recognized identity of the fact to be learned or the action proposed with the growing self; that it lies in the direction of the agent’s own growth, and is, therefore, imperiously demanded, if the agent is to be himself. Let this condition of identification once be secured, and we have neither to appeal to sheer strength of will, nor to occupy ourselves with making things interesting.” Dewey, _Interest and Effort in Education_, p. 7.] _Style_, as presented above, could be rephrased as inevitably _interested_ writing, as a self-exploring and doubly-contingently proceeding alienation: style functions within a system which makes it conform, while at the same time, style _styles_ the system. Understanding it as a sedimented, gradually accomplished kind of parsing, makes it amenable to a processual analysis: unfolding interests reveal a trajectory of _learning_. Interested human writers are the crystals (Deleuze) and filters (Bergson) from which the reduction that is _style_ ensues. The phenomenon of learning as the development of style can itself be said to be the processual link between practical life and theoretical reason, and a goal in itself. “We simply cannot care about what leaves us aesthetically indifferent.” (Brouwer & van Tuinen, _To Mind is to Care_, 2019, p. 6). _Proposals_ of universality and absolute generality (e.g. “artificial,” “intelligence” and “artificial intelligence”), reveal a logic of carelessness at heart (e.g., social indifference, or worse: fascism), which either propose a singular interest as universal, or take interest for granted and are in ignorance of their doing so. _Products_ of universality and absolute generality—we could say: most GenAI—tend to reveal aesthetic indifference and lack of style.^[Or, an abuse and repetition of existing styles.] We will not dive into this last point but merely propose it as an outer edge of attention in our discussion.
However, on top-down mechanisms in pedagogy and/as discipline: if similarity or, by exponential consequence, _absolute generality_ is **not** what we are after, then how to account for the harsh assimilation logic that is fundamental to learning through imitation? This logic can be compared to the observation van Tuinen makes regarding the evolution of the apparatus that was/is the European courtroom: “The court became an artificial world organized around the idea that the impersonality of form—convention, mask, manner, and quotation—provides protection against the chaotic flux of the world and enables composition of different interests and potentials.” (2019, p. 100). Though, however betraying of the pursuit of stylistic singularity—in all things social, pedagogical: creative—there is a functional generality in _likeness_, familiarity or re-cognition, which necessarily drives the organization and _assimilation_ of knowledge, as the term implies. As Tomasello points out: emulative **and** “collaborative learning leads ... to skills in ... **thinking about thinking**, [in this way, children] come to respect rational norms of discourse and argumentation.” (Tomasello 2015, p. 650, our emphasis). Though, while _sameness_, homophily, can regulate the coordination of learning, it should not be its aim: the interesting thing is to _learn_ (think about thinking: metacognitive functions), not to repeat. To remind: these are our fears of a possible paperclip maximizer. Commenting on commonality and citing Bergson, van Tuinen (2019, p. 181, our emphasis in bold):
 
>...each species, each individual even, retains only a certain **impetus** from the universal vital impulsion and tends to use this energy in its own interests. In this consists _adaptation_. ... **Harmony, therefore, does not exist in fact; it exists rather in principle**; I mean that the original impetus is a _common_ impetus, and the higher we ascend the stream of life the more do diverse tendencies appear complementary to each other. ... Harmony, or rather “complementarity”, is revealed only in the mass, in tendencies rather than states. Especially ... harmony is **rather behind us than before**. It is due to an identity of impulsion and not to a common aspiration.’ Bergson, _Creative Evolution,_ 50–1.
 
As Susan Ruddick notes, Pierre Macherey sees something similar, “philosophy ... is not before you as you imagine, but behind you, in that element of the always-already over which the veil of ignorance is usually drawn. ... It is not a problem of not knowing enough philosophy but, in a very particular sense of the word ‘know’ which comprises a relation with non-knowledge, of knowing it already too well, in forms whose confusion needs to be unraveled, necessitating an intervention that will trace within them their lines of demarcation.” (Macherey 2009, in Ruddick 2011, p. 214). What this shows, to us, is that both thinkers are certainly onto the predictive effects of concepts: they are *necessarily* behind us. Homophily is a _bet_ on the possibility that we are similar, not something a “we” emerges from but something which is rather a self-fulfilling tendency. _Interest_ is one of the key markers in language that cracks this predictive effect wide open. Unlike traditional reason, were we are all “in” on the joke and it should therefore be understood, interest immediately denotes its situated and aesthetic dimensions, possible cross-incongruences between perspectives, and sometimes even offers insights into the procedures, or PSR, that preceded its pronunciation.^[“_That_’s interesting”, as an utterance in light of an event, indexing a locus of attention which might invite dialogue or other coordinations of behavior.] In the same vein as the motions of _affect_, as noted by Massumi, interest is not an operation ‘separate’ to reason but actually drives its very capacity to realize itself: it sustains, styles, or indeed _grounds_ its movement. While _reasons_—which work backwards, retroactively, explaining themselves on the basis of some sort of sedimented, habitual logic which is almost always pinched at some edge,^[Like the hairy ball theorem.] always revealing, under rationally, self-alienated analysis: a necessary flaw, an inconsistency—tend to pose as transparent explanatory devices, interests reveal their inconsistency from the get-go. While a _reason_ often pretends to be timeless, inevitable, as if it always had been what it is, an _interest_ reveals itself as contingent and conditioned. Citing Nietzsche: “To make fun of writing but also to make fun with it: an art which implies knowing ‘what can be done with language as such’ (_Ecce Homo_, ‘Why I Write Such Excellent Books’, p. 4), Sarah Kofman writes (1993, pp. 3-5, our emphasis in bold):
 
>Yet despite this subversion of language one still runs the risk of being understood, of being heard, misheard, translated into another language: the game of writing, for Nietzsche, remains subordinate to a new art of interpreting the world, the communication of a new perspective. … by attracting [new followers] on to ‘new secret paths and dancing places’ … But how is it possible to communicate ‘**personal**’ views using a language which, despite the displacements to which it is subjected, remains **common** and vulgarizing? Without ‘speaking badly’, how can one express a Dionysus who speaks a language totally different to that of Schopenhauer or Kant? … Is it not rather a way of acknowledging the **specificity** of philosophy, its irreducibility to any other form of expression—even if this philosophy no longer has anything traditional about it, even if it is **unheard-of** and **insolent** philosophy? A philosophy which, by combining all the ‘genres’ in its writing, deletes all oppositions with one great burst of **laughter**.
 
It is probably clear from our contorted excursions how much this project’s interests line up with those above. The contradiction Kofman presents reveals not only the interested drive of (extreme) style, but also the (docile) interest in remaining dialogically relevant, in coordination with the larger conversation. To refuse the _call to order_ (Moten and Harney), is to abuse and become abused by it, too. This _neither-here-nor-there_, intermediary, hesitant, expectant and explorative condition, can be said to be the very state of learning, of _assimilating_ knowledge, itself. As Deleuze also notes: one can only write about that which one knows badly, about that which one doesn’t yet know (_Difference and Repetition_, 1994 (1968), French preface), there is no other way, despite polished images of static knowledge. Because it exists in this in-between (_inter_-: between, and _esse_: to be): interested **writing**; expression, is a primal example of the _incorporated_ dialogical learning we will refer to throughout this text. Incorporated because the “corpus” of language enters the _human corpus_ and slowly becomes an internal dialogue. As Negarestani puts it in _Intelligence and Spirit_: “[t]here is no process that does not speak of another process ... there is no monologue or private thought without a dialogue, an interaction within and over language, and correspondingly there is no dialogue without an information gain or new knowledge made possible” (2018, p. 357). The learning of this dialogue, as we will see, can be said to emerge by way of others’ previously interested indications; pointings (others’ lives, and how they are interpretably interesting, relevant). Here, though contortions, we are therefore styling the concept of _interest_, not only to say how it’s fundamental to any reasoning, but also to explore the conceptual structure it reveals, which is absolutely everywhere and nowhere at the same time. Honing in on it allows us to reveal schemes of care; mimicry; (trans)individuation; of bias; of systemic limitations, and much more, through an encounter with a concept so colloquially ready at hand.
 
### AI and the permanence of risk
 
>To begin with **learning** machines: an organized system may be said to be one which transforms a certain incoming message into an outgoing message, **according to some principle of transformation**. If this principle of transformation is subject to a certain **criterion of merit of performance**, and if the method of transformation is adjusted so as to tend to **improve** the performance of the system according to this criterion, the system is said to _learn_. A very simple type of system with an easily interpreted criterion of performance is a **game**, to be played according to fixed rules, where the criterion of performance is the successful winning of the game according to these rules. ... In these games, we not only can theoretically find a best policy for the playing of the game, but this policy is known in all its details. The player of such a game (either the first or the second) can always win, or at any rate draw, by following the policy indicated. In theory, any game can be brought to such a state—this is the idea of the late John von Neumann—but once a game has been brought to this state, **it loses all interest**, and **need no longer be considered even as an amusement**.
>Wiener, “God and Golem”, 1964, pp. 14-5, (our emphasis in bold).^[More on John von Neumann and game theory here: [[Games]].]
 
>I’d be surprised if within the next few decades or something we don’t have AI’s that are truly smart in every single way, and that are problem-solvers in almost every single important way. And I’d be surprised if they didn’t realize what we realized a long time ago: that [the rest of the universe contains the materials for continued growth]…. In the beginning they will be fascinated by life, they will want to understand that completely, just like people today want to understand how life works, and the history of our own existence … **Once they understand [life] they will lose interest, like anybody who loses interest in things he understands**. And then as you said, the most **interesting** sources of information for them will be **others** of their own kind. So at least in the long run there seems to be **some sort of protection through lack of interest on the other side**. And now it seems also clear, as far as we understand physics, you need matter and energy to compute and build more robots and infrastructures, and AI civilizations, and AI ecologies … So it seems inconceivable to me that this thing is not going to expand, … it’s going to expand, limited by light speed and physics, but it’s going to expand … so there’s plenty of time to conquer the entire universe.
>
>Jürgen Schmidhuber, 2022, Lex Fridman interview.
 
The two quotes above are presented in order to reveal the self-alienation of reason as interest, through the whims of language: once something is supposedly understood (i.e., has been rendered rational through distributed dialogics of some kind), it should no longer be interesting. It stops evolving, somehow. Reason, language, or intelligence, only wants what it cannot have. Or, a similar thought elsewhere: Nick Bostrom, discussing the simulation argument^[See: [interview on YT](https://www.youtube.com/watch?v=nnl6nY8YKHs) (2013).] presents the idea of the possible “loss of interest” in running ancestor simulations. Interest is mentioned countless of times during this interview, often using it to frame the argument that if humanity, as we know it, reaches “technological maturity” and is at that point “still interested” in creating ancestor simulations, then we can actually estimate with high probability that we are, currently, in a simulation. Early interests in cybernetics seem to line up with current AI interests. The function of interest seems to preserve itself across generations.
In terms of interest, the concept of artificial intelligence _seems_^[More on this later.] to play a doubly-transcendental role: on the one hand the concept (re)presents itself in infinite postponement, as that which humanity has yet not achieved,^[I.e., the infamous “AI effect”: the perpetually moving of mileposts.] providing both an interesting semantic attractor for the pursuit of a total _other_ outside the human,^[Which is nevertheless a servant to the human, and thus _similar yet docile_: AI should be able to perform tasks (in a selected few’s interests).] as well as a suspension of (practical) judgment (resulting in, among other things, the apocalyptic race towards “AI ethics” or guidelines resounding across institutional halls).^[The implication through our observation being that because since AI is presented as omnipotent but also as always just beyond reach, deliberation on it means that practical considerations become both urgent yet perpetually preliminary or anticipatory. Rather than rendering concrete judgments of AI as it actually exists today, we are eager to create unrealistic frameworks for a technology we define as not-yet-fully-realized. This creates a situation where ethics becomes more about preparing for an imagined future AI than engaging with the realities of current AI systems and their impacts. One of the obvious “social dissonance” (Mattin 2022) blunders here is military AI, especially in the recent context of big tech’s involvement in Gaza.] On the other hand it brings to light the _problem_ of transcendence itself: confrontation with absolute limits (of interest, knowledge, capacities, futurity, etc.).
 
>I believe Émile Borel pointed out, it would be better to generate just the interesting theorems, not all the theorems; most of them are totally uninteresting! But no one knows how to do that. In fact, it’s not even clear what an “interesting” theorem is.^[Chaitin points to Wolfram, as having an inkling about “what makes theorems interesting in his book [A New Kind of Science, 2002]. As usual, he studies a large number of examples and extracts the interesting features.” What Chaitin means here, is he extracts the things that appear to break rules: automata that don’t halt at clear patterns, complexities that keep on growing.]
>
>Chaitin 2004, p. 29.
 
The life of these phenomena, we argue throughout this project, is therefore renewably conceptual, and _permanent_ semantic transformation at the very limits of thought is its most prominent “feature,” certainly not a “bug”.^[Unless the _bug_ is noise, an interesting parasite (Serres). This is treated in [[03 Semantic noise]].] Knowledge _is_ repeatable in many ways. Look at these words. But the _image of all knowledge_ as repeatable inherits too much from an industrial image of ‘useful’ mechanical reproduction.^[And, later of course, from software as separate from hardware, as Kittler criticized (2014).] Interesting repetition is always the result of mutational surprise. Function prevails over form, there is something to putting the world, itself, to _work_:^[Quite literally in ML, as Rieder notes, citing Domingos: “The assessment that machine ‘[l]earning is more like farming, which lets nature do most of the work’ (Domingos, 2012, p. 81) is clearly incomplete. Even if it is certainly not wrong to say that ‘[m]achine learning systems automatically learn programs from data’ (Domingos, 2012, p. 81), this ‘program’ – or decision model – sits among a whole range of other programs, that is, things that are not ‘nature’ or ‘the world’.” (Rieder 2020, p. 250).] the _workable_ (_wirklich_) is rational.^[This is explained in [[10 Bias, or Falling into Place]].] From the person having become a part of the machine (Marx), to the machine having to become a person.^[This is the issue of homophily and behavioral contamination we treat in [[05 Prediction]].] This is no fetishization of aleatory uniqueness or artisanality, though, please mind our words. We are simply trying to relax and be pragmatic: look at **interest** here, again, where is its function going?
Interest reveals a thing or two about what is at stake in the (proposed-assumed) transcendence of AI, inheriting from this image of reproducibility.^[Which, as we treat in [[05 Prediction]], might just be a biological consequence.] The simple and very first excavation layer of this revelation is the realization that “bias” (i.e., interested style) overturns any possibility of transcendence: it is always concrete. A second layer might be—if _bias_ is an unavoidable feature—that we _return_ to transcendence as a sociorational pointer of predictive indexicality; of communicative access, because we realize that providing an ultimate or total account of diverging interests is impossible (and not only because we are perspectival; intransparent; finite; intractable, etc.). A third, and by no means final, layer of excavation becomes the historicomaterialist one, in light of the first two layers: how to plan and realize technology—as institutions, as AI, etc.—if we cannot settle on what we mean by goals or orientation? Or are these simple images confusing representation for process? In other words: is learning, communicating, or action possible at all without (representational) coercion?
Collective orientation implies a complex kind of possibilistic prediction,^[The best of all possible worlds is only visible from above, thinking of Leibniz.] and the (AI) tools we currently have render a specific form of imagination that collapses possible futures into interested variations on the past. As van Tuinen notes of Bergson on the _possible_: things do not happen because they _were_ possible, they _become possible_ through the unfolding of interpretation: all _apparent_ function evolves. Van Tuinen: “as Bergson argues, possibility means only an ‘absence of hindrance’, which the human intellect retrospectively turns into ‘pre-existence under the form of the idea’: ‘For the possible is only the real with the addition of an act of mind which throws its image back into the past, once it has been enacted.’ Only when the past no longer has an immediate interest can it appear as divided from the present, as if we are dealing with two immobile positions outside time. But this is a projection post factum, since in fact, the present endures in continuity with a virtual past that preserves itself in the indivisible movement of the creative act: ‘It is the real which makes itself possible, and not the possible which becomes real.’” (2019, p. 111). The “absence of hindrance” is what we could criticize as predictive flattening, as anti-evolution. And, important to note: while we use the notion of *possibilistic* often in this project to denote _tact_ for the virtual, this idea of possible tact can be a flattening problem, too.^[Our possibilism is what renders this possible: “In reality and hence in art, by contrast, the possible is not less but more than the real. It is virtual and this means that it requires creative thought.” (van Tuinen 2019, p. 113).] In any case, the point here is to point to how concepts retroactively affirm themselves, leading to the above-mentioned problems of coercion.
We can also frame the above through more concrete contemporary lenses: “algorithmic techniques [can] constitute a domain of research _after_ their inception; [as] they introduce complex mechanisms into already complex settings, and even fully deterministic principles can yield results that are not easily accounted for.” (Rieder 2020, p. 220). In the context of an analysis of the Bayes classifier—a fundamental statistical technique in modern machine learning and an algorithmic reality in anything from filters in email to relevance rankings in digital queries—Rieder presents the concept of _interested learning_ to advance the idea that “contemporary information ordering represents an epistemological practice that can be described and analyzed as ‘interested reading of reality’, a particular kind of inductive empiricism.” (2020, p. 17). The Bayes classifier is conceptually driven by, as well as represents, an ‘inductive empiricism,’ because it implies the “learning” of patterns by probabilistic insights, which inevitably results in the conceptual suspension of _specific_ judgments, and thus renders possible an intuition of transcendentality, while remaining practical and concrete. In statistics something doesn’t have to be, it _could_ be: Rieder quotes M. E. Maron, “a physicist turned analytical philosopher and cybernetician” in the late 1950s. He was the auspice of the representation of indexed information in terms of “degrees of aboutness”, as opposed to Boolean indexing (Rieder, 2020, p. 237), which renders _specific_ outputs; absolutes, we might say. With “degrees of aboutness” we are able to suspend giving into absolutes, but the spectrum which they represent upholds these categories as true, as ideals, in a traditionally transcendentalist essence-versus-appearance, “rule” versus “mere example” manner.
 
>Moten and Harney (_All Incomplete_, p. 165) cite Zukav citing Heisenberg: “[Probability] meant a tendency for something. It was a quantitative version of the old concept of potentia in Aristotelian philosophy. It introduced something standing in the middle between the idea of an event and the actual event, a strange kind of physical reality just in the middle between possibility and reality.”
 
Dialogical learning is not all statistics,^[Well, this can be argued.] but so far machine learning _seems_ to be. If statistics is data with context; interested, then machine learning is the interested contextualization of statistics in novel domains. _Hyperdimensional interests_^[_Fini-unlimited_, as Rieder says citing Deleuze’s concept. This would be: promise of combinatorial infinity out of a finite set. “The calculative power afforded by the possibility to freely apply mathematical techniques to both numerical and nonnumerical information creates a space of experimentation where even highly simplistic representations of language such as word frequency counts can yield spectacular results, at least in well-controlled experimental settings. And the potential for generalizability seems almost limitless.” (Rieder 2020, p. 220).] with absolutist tendencies.^[“Depending on the actual collection and the specific way language has been formalized, transformed, and filtered (Single word frequencies? Co- occurrences? Lemmatization? Dictionary enrichment? Frequency cutoffs? Term weighting?), clusters of items may or may not reflect cultural categories such as topics, and ‘fiddling around’ with parameters is a common practice in areas where the constitution of outputs can be hard to anticipate and even harder to explain.” (ibid., p. 227).] The representation of something as ‘neither here nor there’ in probability can be understood to promote a transcendental image of the concepts under probabilistic inspection as absolutes: they _exist_, but need never be observed with 100% certainty. This technical imagination renders contemporary information-processing and ordering as the _promise_ of knowledge, a knowledge which is of course always retroactively co-evolving with its contextual enculturation. In the case of the Bayes classifier, already-existing categories of knowledge, based on enculturated patterns are, of course, an _absolute_ necessity when it comes to the production of (new) insights from unordered data. In-depth analyses looking into the interested histories and the effects of these technical implementations, such as the one Rieder proposes, can render our absolutes as rather _malleable_ (and therefore plurifunctional), by attending to “how software-makers embedded in concrete circumstances think through or with the machine” (ibid., p. 243).^[Hopefully resulting in at least some “rage alongside the machine” (van Tuinen, 2021, n. pag.).]
Concretizing conditions of interest reveal “... the ‘optics’ of machine learning [as] always ‘partial’^[McKenzie 2017, p. 80, cited in original] in both senses of the word: partial as incomplete or fragmentary; and partial as skewed, biased, or—preferably—**interested**.” (ibid., p. 255, our emphasis).^[“If one considers technological artefacts from the point of concretization, then, one may better see each artefact in light of a technical object’s virtual schemes. These leave open a space for contemplation of technical evolution as well as space for other considerations” (Iliadis 2015, p. 94).] If we follow the classic “we do not understand it unless we can program it”,^[A classic technical dictum, one often promoted by figures like Wolfram or Chaitin.] we certainly follow the idea that we ought to already know what it is that we want before we build it, the tool needs to reveal a function. In the processes of concretization of tools (Simondon), this is where we run into all of their frictions. But saying that only constructing and seeing _function_ is equal to “understanding” means there is little space for function-evolution: we already have an idea on how to technically-grip reality, and implementation is a side effect. According to Rieder it also leads to the tautology that if: “a technical idea is [only] true if one can build it, [then] if one cannot build it, it is not a technical idea and therefore has no merit.” (ibid., p. 27). Understanding the linguistic unfolding that this occurs through, how it is pedagogically approached; how the evolution of functionality is a path riddled with—or composed _only of_ metaphors—can show us much about the interests enculturated in technicity (this is strongly argued by Agre 1997, from whom Rieder also takes inspiration). What does _knowing_ this do with our image of _reproducible_ knowledge? In the context of learning (through) language: the ways in which language can be (meta)learned _as_ language^[That is: all language is speculative because it is our collective possibilistic grip on reality-modulation, as we will see later on with Sellars’ _looks talk_.] and therefore become stylized, interestingly modulated, reveals some openings, as we will see.
In the context of a possible _Poltergeist_ in the machine, and the desire for things like explainability or complex risk-assessment; the desire for non-human agents to give us a principle of sufficient reason, we need to contend with interest _before_ explainability since—computational time allowing—we’d want elucidations of choice-sequences _before_ agents engage in complex actions. The unfolding of this becomes a rather interesting domain in the context of _chain of thought reasoning_, where LLMs are probed for their inner logic by asking them to explain the steps by which they arrive at answers.^[“[N]on-trivial reasoning problems require multiple inference steps. LLMs can be effectively applied to multi-step reasoning, without further training, thanks to clever prompt engineering. In chain-of-thought prompting, for example, a prompt prefix is submitted to the model, before the user’s query, containing a few examples of multi-step reasoning, with all the intermediate steps explicitly spelled out (Nye et al., 2021; Wei et al., 2022). Including a prompt prefix in the chain-of-thought style encourages the model to generate follow-on sequences in the same style, which is to say comprising a series of explicit reasoning steps that lead to the final answer. As usual, the question really being posed to the model is of the form “Given the statistical distribution of words in the public corpus, what words are likely to follow the sequence S”, where in this case the sequence S is the chain-of-thought prompt prefix plus the user’s query. The sequences of tokens that are most likely to follow S will have a similar form to sequences found in the prompt prefix, which is to say they will include multiple steps of reasoning, so these are what the model generates.” Murray Shanahan, “Talking about large language models” (2023, p. 8).] GenAI systems are often charged with “hallucination” as a side-effect, which seems rather naïve if we think from a perspective that all interconceptual (Wolfram 2023) information on offer through these hallucinations is the possibility of _something_. It might not be what _we_ are interested in, but it is a determinate, **real** path between structures. The rather hallucinatory part, and what reveals humans as nichily-limited and highly interested readers of these systems, is that we assume there is an underlying PSR, for _us_. Often, there is not, but only from a human perspective. As Rieder mentions, what is important is that the “ever-expanding capacity to read and assess the world in relation to a **purpose** (Rieder 2016)” (Rieder, 2020, p. 251, our emphasis) is given more attention. Techniques such as the Bayes classifier are, inevitably, “devices for the automated production of _interested readings of empirical reality_” (ibid., p. 252, emphasis in original), which drives our analysis into _interest_: without taking the concept for granted, but digging into the interests of interest. It might also go without saying, in the context of this project, that the notion of interest is very useful for considering frame- or attention-type problems in AI, as well as the historically corresponding paradigm of _personalization_^[Filtering, echoing, encapsulating, isolating.] in digital environments.
Additionally—as covered in the chapter on semantic noise—if the complex recombination of probability distributions is what has come to determine the fluid proficiency of language models, should we not challenge the simplifying generalist tendencies in the _interested rationale_ underlying their designs and intended purpose? These systems are more than their engineering gives them credit for. The answers, for this project, lie in an analysis of _becomings_,^[Famously, (Bergson-inspired) becomings, for Deleuze (and Guattari, 1987), are anti-representational in that they are anti-mimicry, imitation.] through the tortuous transformation of concepts in an attempt to witness them _as_ indexical things, with the capacity to predictively aim our inclinations, but also as systems evolving to their own accord, _through_ human parlance. And, additionally, in proposing a slower contemplation of the immense scope that is the landscape of probability distributions in current language systems: something far bigger than anything we can get cognitive grip on,^[“One can view the project of AGI through [the lens of] humans enacting expressive bootstrapping via models which perform feats entirely incommensurable to human reasoning, leveraging internal languages which increasingly do not yield semantically to human observation” (Cavia 2022, p. 99).] and perhaps the reason we predictively resort to intuitions of generality and transcendence, inevitably, as they simplify the future by making it _flat_.
As Rieder reminds us in the context of the Bayes classifier and our current condition: “[w]e no longer decide based on what we know; we know **based on the decision we want to make**.” (2020, p. 256). A reversal which can render (language, knowledge and) reason as only capable of utilitarian exploitations, less so of interesting explorations. “[T]he product of poesis-engineering is not just a model in the traditional sense of representing something [it is] to reframe what is given to us at different scales or joints^[Note from S. de Jager: chunks or parsings.]: ... cutting at them by means of _technical_ and _epistemic_ strategies so as to progressively find subtler joints at which reality can be revised and reconfigured—that is the task ahead.” (Negarestani, 2021, n. pag.).^[“What does it take to make anything at all? Plato’s craftsmanship as the blueprint of poesis-engineering”: https://miguelabreugallery.com/the-poet-engineers-reader/#rezanegarestani. Accessed June 2022.] This is a conceptual task, a linguistic condition, hardly a phenomenon restricted to AI. It bottoms out at a political chunking and parsing matter of who gets to live, and how. Interesting style, the evolution of creativity, is a matter of randomness naturally inducing itself into stubborn (boring) games of who gets to live and how. _Interest_ always seems to point to this: the sedimentation of new patterns results from recursive entropic surprisal, and the allocation of _attention_ to this phenomenon. “The path toward scientific truth is always a computation of computation, a thinking of thinking, an observation of observation.” (Hui 2019, p. 179). Contemplations on this now are what will determine our possibilities of “coexistence” (ibid, p. 131) in the future.
 
### Complex interests encoded in simple words: a _metareading_
In the essay “The myth of the framework” (1976), Karl Popper, truly _hating_ on any kind of (possibly infinitely-regressing) relativism—which, for him, leads to accepting incommensurability between peoples, barring communication^[We dissent, as we think a generous take on relativism opens up perspectivisms, it really depends on the—_sorry_—framework we are _given_ to.]—reflects on the following (recursive) effect of language: all languages clearly evolve, they change to express _new_ ideas, and comparing languages by translating between them gives us more new ideas. Therefore no two _frameworks_ can be incommensurable: translation must be possible if we are talking about the “same” reality. “Rational cultures” evolve by “clashing” and adjusting to each other, says Popper (p. 39). We opt for a similar presentation of this phenomenon, but in _this_ framework:
 
>Anthropology, then, is interested in equivocations in the “literal” sense: _inter esse_, betweenness, existing among. But, as Roy Wagner said of his initial time with the Daribi of New Guinea (1981: 20), “their misunderstanding of me was not the same as my misunderstanding them,” (**which may very well be the best definition of culture ever proposed**). ... **Equivocation is not error, deception, or falsehood** but the very foundation of the relation implicating it, which is always a relation with exteriority. Deception or error, rather, can be defined as something peculiar to a particular language game, **while equivocation is what happens in the interval between different language games**.
>
>Viveiros de Castro, _Cannibal Metaphysics_ (2009) 2014, p. 90, our emphasis in bold.
 
_Equi-vocation_, originally denoting the idea of “same voice”, has come to represent error, fallacy or indeed _clash_. But as we will see in our treatment of deception and lying later, this effect—as Popper noted: how languages change—is what reveals **the interest of language to remain a thing its creatures have to adapt to**, it is how language guarantees its survival. It is the “genetic pool-mixing” of language, through creatures. Redundancy in language, which is what leads to what _appear_ as rules from our limited vantage point, reveals the persistence of _function_ in the face of contingency. Language is able to remain an evolving system, sustained by creatures such as humans. We take William Burroughs’ proposal of language as a virus (1967) very seriously here.^[“Burroughs claims that _The word is now a virus_. The flu virus may have once been a healthy lung cell. It is now a parasitic organism that invades and damages the central nervous system. Modern man has lost the option of silence. Try halting sub-vocal speech. Try to achieve even ten seconds of inner silence. You will encounter a resisting organism that forces you to talk. That organism is the word. (49)” Gontarski 2020, p. 555.] If it were merely a tool or a game, _for humans_, language would indeed be unequivocal, indexical, translatable, a matter of _rules_, and would not be language as we know it: i.e., a dynamic evolutive process with future _tendencies_ that is able to transfer across creatures of the same and of different species, it is not a rigid rule-based representational model effectuating step-wise.
Popper comes, allegedly, in peace: proposing incommensurability is dangerous for _people_, we agree. But there are different ways to understand this incommensurability. His title is riffing off of the famous “Myth of the given”, the problem of naïve realism, approached by Wilfrid Sellars, who also challenged a lot what humans supposedly do in and out of different languages, and in the interested interstices between these supposedly separate modes.^[A separation which we do not uphold here.] Sellars (1954b) criticizes Popper’s understanding of language as largely _descriptive_, as this fails to see how language “games” do much more than that. We will play Sellars the same card he plays Popper, and say: languages cannot be described as rule-bound games. Language, in general, as a large-scale evolving structure now finding its way through/as technological extensions more and more, is very much _playing us_.^[The definition of language given in the introduction of this project is, to remind: “**Language**: the mark of _any_ language is the establishment of an arbitrary bond, which results in the preservation of that arbitrariness as a function. In general, the ensuing effect of this, leads to language as something that can be described as a system dynamically sustained by a collection of beings, which supports-controls the collective discovery-invention of affordances. Language can be understood as a life of its own, having the capacity to modulate predictive attention _through_ creatures. Language is what _enables_ a lot of chunking and parsing, and therefore gives us _meta_-physics. Language’s redundancies can, if observed; chunked, for periods long enough, be understood in analogical parallel with the lives of creatures: each word, like each individual, is an attempt which serves for the survival of an (initially always arbitrary and aleatorily-produced) function, a trait.”] It has an interest in survival, and therefore has to feed on contingency (the mutating effects of which, to limited human beings, reveal themselves as error, equivocation, lying, etc.).
According to Sellars, episodes of thought are ““in” language-using animals as molecular impacts are “in” gases, not as “ghosts” are in “machines”” (cited in Brassier 2007, p. 6). Thoughts are thus _inevitable_ in language: thoughts happen _to_ language.^[And language happens to creatures.] In this project, however, we do believe they are like ghosts in machines: even though we’ve never seen a ghost (language makes us say this).^[I cannot but press on this: by deeply inspecting _any_ word we submerge in endless pits of supposed pure nonsense. Yet we still believe.] To exemplify, however, the animation of your physiology by way of these scribbles is what we would call such a ghost: the equi-voice reading this in your experience. Sellars’ famous critique of the _myth_ of a supposed givenness which fundaments thought and parlance, is the critique of a position which doesn’t understand its own implication in the conceptual matters at hand: all access, all perception is (self-)mediated and conceptually corrupted (Kant). What Sellars insists upon is that conceptual thought is inferential, and it is, at root, linguistic, and language is, at root, a normative matter (famously presented by him as *the game of giving and asking for reasons*).^[“I take intentionality… to be the mark of the mental. I agree with the classical view that there is a domain of ‘inner episodes’, properly referred to as ‘thoughts’, which are not linguistic—though they are analogous in important respects, syntactic and semantic, to linguistic structures, and are functionally connected with linguistic behavior. (Sellars, W. (2007). “Mental Events.” In _In The Space of Reasons: Selected Essays of Wilfrid Sellars_, p. 283)” Cited in Cavia 2024, p. 4.] As Cavia notes, however: “when a salamander extends its tongue to catch a prey which is in motion, it is exhibiting capacities of recognition and prediction ... functionally linked to its **intention** to eat, all **without recourse to language** ... [from Sellars’ position] there is no convincing account of how this position can be paired with inferential semantics to allow us to distinguish between, for example, rat and human cognition, without flattening the field of non-human subjectivity.” (Cavia 2024, p. 5, our emphasis in bold). The result of this tension is, therefore, an unresolved boundary problem in Sellars’ epistemology: his critique of the given dismantles naïve realist-empiricism, but its organizing principle leaves one wondering about what conceptual resources language-users need in order to articulate the (dis)continuities between different perspectives. All perspectives clearly latch on to and are able to (creatively) chunk reality. Additionally, Sellars’ account presents language-interested humans as _users_ of language, we argue for the opposite: language uses us.
 
>I presume that no philosopher who has attacked the philosophical idea of givenness or, to use the Hegelian term, immediacy has intended to deny that there is a difference between _inferring_ that something is the case and, for example, _seeing_ it to be the case. If the term “given” referred merely to what is observed as being observed, or, perhaps, to a proper subset of the things we are said to determine by observation, the existence of “data” would be as noncontroversial as the existence of philosophical perplexities. But, of course, this just is not so. The phrase “the given” as a piece of professional --epistemological -- shoptalk carries a substantial theoretical commitment, and one can deny that there are “data” or that anything is, in this sense, “given” without flying in the face of reason.
>
>Sellars, 1956, p. 127.
 
The quote above, to us proves the point of how concepts are _all_ actively inferential. The point Sellars wants to make is to deny the possible philosophical-scientific purchase of any notion of _givenness_. No such ‘direct line’ with reality is possible, the very way language engages reality reveals this quite clearly: when we speak of things _seeming_, _looking_ _like_, or _appearing_, we are in effect inevitably saying that they could be otherwise, that the underlying reality may not be as we naïvely perceive it. The _manifest_ image where these language games take place, plays a regulative role for Sellars, “in the sense that it provides the framework ‘in which we think of one another as sharing the community intentions which provides the ambience of principles and standards’” (Sellars 1963 cited in Brassier 2007, p. 5). This ambience is how language contains and coerces. Sellars’ project was to harmonize the apparent disjunction between the manifest and scientific images. As different ways to access reality, these two are dialectically subject to each other: the scientific image is able to reveal aspects of the structure of reality,^[We would say, the revelation of new _functions_.] and the manifest image bottoms out at irreducible behavioral norms that define what it is, for Sellars, to be a member of a rational community. His belief was that manifest, psychosocial “looks talk”^[When we say something “looks like” or _seems_ like (as we will explore) something, we mean that there’s a filter, that we know things can _be_ a way but _appear_ in other ways.] could be scientifically naturalized, rendering a “stereoscopic” image, perhaps inducing new dimensions of thought/life/meaning.
As Brassier puts it, the manifest image “indexes the community of rational agents ... [where] ... the primary component ... is the notion of _persons_ as loci of intentional agency” (ibid.). Unlike the scientific image, which presents a _different_ framework of epistemic orientation, one lacking persons, interests and, therefore, promoting a certain indifference, the _man_-ifest image is what makes _persons_ out of persons (or man/men, says Sellars). That’s how it is normative: it creates the field of social possibilities, by showing what is possible _for_ people, through the ins and outs of natural language. The chunking here exposes interaction as supposedly _volitional_, agency as _situated_, purpose as _oriented_; interested, and of course _limited_. Brassier notes how “echoing Kant, [Sellars] concludes that we have no option but to insist that the manifest image enjoys a _practical_, if not theoretical, priority over the scientific image, since it provides the source for the norm of rational purposiveness, which we cannot do without.” (ibid., p. 6). As mentioned, what Sellars wants is to _integrate_ the manifest (interested, social and limited) and the scientific (detached, impersonal and possibly unlimited as it is an ever-self-correcting affair) images. What we want to explore is how (meta-)language learning can give us an insight into the disorientations this might entail.
The essay “Some reflections on language games” (1954) by Sellars is a text dealing with a _possibly_ disinterested notion of learning how to observe the (rational) rules that guide language-learning. Sellars presents the _apparent_ paradox that emerges when we think about languages and the learning of their rules: in order to grasp the indexical ‘rule’ that, e.g., words _represent_ concepts, we first need said possible relationship—_representation_—as a transcendental concept, for which we might need a meta-transcendental concept, and therefore another meta-concept, _ad infinitum_, regressing infinitely. His suggestion for a bootstrapping solution is that we should understand language operations as analogous to game operations: we entertain a Kantian position of _as if_, and perceive the language system we are **in** from a perspective where we simply play along—“conform” to “patterns” (p. 204)—before we understand what we are **really** doing. This, we could say, is akin to Nietzschean lying at its best, and forgets about “language [a]s fascist” (Barthes), or about the inherent violence in meaning (Derrida, Spivak) at its worst.^[See notes on [[Misosophy]] for more on this subject.] For us, it is also evidence to how language, as a kind of substrate-independent software, has humans much more than they have _it_. Thoughts are, then, very much like “ghosts” are in “machines” (Sellars cited in Brassier 2007, p. 6). Thoughts, like William James said, are _thinkers_, themselves.^[This is often cited by Michael Levin, too.]
In order to reflect on how this permanent struggle of “vicious regress” (Sellars, p. 204) is actually the normal (predatory, parasitic) effect of language on its substrate, an effect which is able to render it an evolving structure, we will follow the first few sentences of Sellars’ essay by reflecting on the words he employs **_as if_ they represent the concepts they lay literal claim to**, from our limited point of view (i.e., we will try to follow the game by the rules, and not let ourselves be guided or _conform_). This is precisely what Sellars does with Popper, too (1954b), we therefore invoke a ghost of mimicry’s past. The _interested_ reading we will explore—besides being obnoxious and tireless—goes in line with our observations in [[04 Concepts as pre-dictions]], where concepts are semantic attractors which estimate possible futures, or perhaps in the intuitionistic computational logic which Cavia (2024) explores around Sellars’ inferentialist semantics: the philosophical move here being the treating words not _just_ as assigned types, but as (multiply-)realizable programs that, when executed, produce specific, interested realities (which can be organized as types, for Cavia, we will not explore this here). This is thus a reading “against the grain” of the first few premises of the essay, which begins the following words:
 
>It seems plausible to say that a language is a system of expressions the use of which is subject to certain rules. It would seem, thus, that learning to use a language is learning to obey the rules for the use of its expressions. However, taken as it stands, this thesis is subject to an obvious and devastating refutation.
>
>“Some reflections on language games”. Sellars, W., 1954, _Philosophy of Science_, 21(3), 204-228, p. 205.
 
“It _seems_”—immediately the text presents the reader with the dilemma of “looks talk” (Sellars), of appearances and their ontological characterizations, as well as with the tacit back-_ground_ assumed by Sellars as one which should be immediately acceptable to a reader. Cavia (2024) also notes that Sellars’ inferentialist semantics aim “is to bring nominalism to bear on the meaning of ‘means’ itself in [a] formulation” (Cavia 2024, p. 12). We contrast this representationalist, indexical, illuminated verb (i.e., to _mean_^[Originally stemming from _medium_, in the middle, like _interest_.]) with the ‘softer’ one Sellars employs himself: to _seem_.^[Perhaps in line with Cavia, who explores a computational interpretation of Sellars’ formulation through the realizability of types (e.g., “triangular(ity)” is a type that applies to many things; we can construct things out of this type as a _program_) as processes, we could present—following Sellars’ proposal that “looks talk” reveals the non-naïvety inherent in language—_seeming_ as verb with _estimating_ functionality, where some proposition P(x) maps onto a probability space [0,1]. However, this misses something crucial about “seeming” and how interested language users are subject to it. It seems it’s not just about probability but about the evidence _pathway_, where dialogically negotiated, retroactive reasons preserve some kind of _function_. The function being preserved would be the relationship between evidence types and conclusions, perhaps a hyperprior of “cause and effect”: even as justifications are retroactively adjusted (because of new evidence, changes in framework) the fundamental relationship between evidence and conclusion, the PSR, remains.] Something _seems_, publicly—i.e., connecting two or more perspectives, as in the case of this essay—when _we_ are able to speculate about what _we_ think others perceive and think about (this is what Sellars says, we agree). Sticking to the context in which the word is used here and not attempting to create formal closure of the entire semantic landscape where “seeming” apparently dwells, as far as this _seeming_ is concerned: it is inviting others into a predictive proposition, by the words that follow. Whatever it is that “seems” is only explained _after_ we are already invited to imagine ourselves imagining it. This is the lesser learning or coercion dilemma, the fact that we need to follow something we don’t know the direction or outcome of: we simply go with along the text. The larger coercion dilemma is that the decision has already been made for us: it _seems_ language can be described like this or that, and that seeming is **wrong**. Different levels of tacit interest can be dissected if we continue down this line, but let us continue reading Sellars’ sentence. (Chunking) time is of the essence here, what pertains interest, learning and reason, as will become apparent through what follows in the section after this one.
“It seems _plausible_…” The characterization of this ‘seeming’ as _plausible_ goes a specific step further in cushioning the democratizing, probabilistic blow. “Plausible” is perhaps preferred over _reasonable_ or _probable_, because to say “reasonable” would mean granting the _seeming_ a level of groundedness which is actually not there—we are told, if we follow a few lines further and read that this seeming, plausible thesis is “subject to an obvious and devastating refutation”. _Probable_ would mean the proposition could, in some future instance, be true, which as we were just made to follow: it is not.^[We’re doomed from “it” already, but let’s focus on “seems”.] It would seem plausible to say that presenting a straw-man argument or proposition, a refutable lie such as this one, is a fair way to begin another proposition. As such, whatever something _seems_: what follows is _not_ that, because of an additional (to be revealed) premise. Following this path of negation, which we are _made_ to follow, let’s follow along a few more words: “It seems plausible to say _that_”. To say “to say” is interesting. Trying to conceptualize what came before, this sentence is proposing, by way of a seeming plausibility, that something can be proposed (by others). A pronouncement can indeed take place, no doubt about it: it is hereby being pronounced. But that pronouncement is wrong, because it is merely _plausible_. It isn’t even, it _seems_ plausible.
Similarly to Sellars’ “looks talk”, Isabel Stengers has also remarked on the use of the emphatic “really”—which we can compare here to “seems”—that whoever pronounces it inevitably marks an ontological divide between the truly objective and the merely apparent.^[She speaks of animists, in her context: “Coming back for a moment to the anthropologist Philippe Descola’s classification, I would guess that those who are categorized as animists have no word for “really,” for insisting that they are right and others are victims of illusions. (2012, p. 4).] And by the way, plausible is just interesting enough to receive collective consenting affirmation, an _applause_. _Plausible_, as an adjective dating back to its 1540s use, denotes something being “acceptable, agreeable; deserving applause or approval”.^[More thoughts on this in [[Grue-Bleen]].] Stemming from _plausibilis_, from _plaus_-, past-participle stem of _plaudere_ “to applaud”. From 1560s onwards we get the sense of: having **superficial appearance** of trustworthiness. It is here that we get closer to the use in the sentence we’re focusing on: it looks like truth, but it isn’t (but how does it _look_ like truth, then?). It seems to seem, to seem, to seem, etc. Again: infinite regress. “It seems plausible to say that…” i.e., it looks like it looks like a statement can be made, but, if we follow along, this statement cannot be made. But, somehow, it seems that way. This goes in line with Sellars’ rationalistic _as if_ interest: in order to produce a different kind of access to reality, thoughts need to do more than just _appear_ to think. But do they? “Appearing” to think being a manifestation of these seeming, plausible standards which endure simply because they endure: conformism, tradition, habit, etc., which would lead to a lack of creative generativity; of rule-creation.^[The etymology of the word _seem_ is equally interesting: at its earliest it stems from the Old Norse _soema_ “to honor; to put up with; to conform to (the world, etc.),” from the Proto-Germanic _somiz_ (Old English _som_ “agreement, reconciliation”) and the sense of “appear to oneself, think oneself” is from 1630s. Additionally, in Middle English it also denoted “to present oneself, appear; be visible, be apparent” (late 14c.). Source: etymonline.com, accessed March 2021.]
“It seems plausible to say that a language is a system of expressions the use of which is subject to certain rules.” “A system of expressions” needs to be expressed, as well as “subject to certain rules” because simply saying “a language is a system of expressions” might indicate this system is less rule-bound than Sellars is interested in. However, a _system_ is (sort of)^[We use these kinds of softeners on purpose and for humor here.] by definition an organized whole based on a set of correlated principles; i.e., rules. We can, however, accept that some things we call “systems”—such as languages—are chaotic, open-ended, and/or irrepresentable enough to elude the strict formalization of their underlying mechanisms, especially when we know they mutate so quickly. Which reveals this tautological instance as doubling down, precisely in order to remark that there is a—seemingly—_high degree of specificity_ to the rules that form a language system. However, we should certainly be wary of the word “certain” in there: _certain_, denotes, both literally and etymologically, the fact that something can be _known_ for sure, with certainty, and more contemporarily, like many other comparable words that put things on hold in the manner of illusions,^[Further thoughts on this are presented in [[04 Concepts as pre-dictions]], [[05 Prediction]], or with [[Noise]] and [[E Pointing]] as forms of [[Representation]].] it denotes that there’s a degree of texture, depth, complexity or _knowability_ that is yet to be determined. We don’t have time to enumerate the ways in which something is in its totality, hence the metaphysical effects of words like _certainty_ or indeed: _totality_: they tend towards _possibility_, never becoming accomplished.^[_The absolute is lazy_, see: [[10 Bias, or Falling into Place]].] Additionally, because we’re able to halt, as limited humans, let us just say, for now, it’s “certain.” Let us say, then, that saying both “system” and “certain rules” is only _a bit_ tautological (which is interesting, as it puts the tautology on hold, for a bit).
We shouldn’t forget to say something about “the use of which.” A system of expressions bound to certain rules, used _for_ expressing something, _by_ someone. The words continue: “It would seem, thus, that learning to use a language is learning to obey the rules for the use of its expressions.” It continues to seem, this _seeming_ extends from the first one, a truth is only apparently certain: it is unprincipled. Sellars continues, “However, taken as it stands, this thesis”—the thesis that language is a system of expressions, the rules of which we learn to obey—“is subject to an obvious and devastating refutation.” Why? Because, as Sellars insists throughout the essay: “learning to obey the rules presupposes the ability to use the metalanguage (ML) in which the rules for [Language] are formulated.” Again, Sellars thinks this is problematic because it leads to an infinite regress: a metalanguage assumes knowledge of a meta-metalanguage, etc. Every time we should exist in a new outside position to have _ruling_ perspective on the previous one, etc. To sum up and round off this metareading exercise: Sellars’ solution would be that _contra_ this mere and apparent simple rule-conforming language-learning: it is through mimicry and corrective feedback—i.e., we would say, through the givenness of coercive discipline—that a language learner first _conforms_ to patterns and then _learns_ to gain perspectival agency to act “freely” within a system of rules. To us, however, there are _only_ patterns, and what percolates _through_ us—chunking and parsing creatures which latch onto these patterns—is an ever-changing function which seeks effectuation _elsewhere_ in order to survive because its substrate is unreliable. This is also argued by Levin (2024) in the context of thoughts or memories: memories being revelatory of how creatures deal with change by preserving function over substrate.^[On perspectivism, he also says: “The capacity for creative robustness is implemented by the polycomputing property—the ability to see the same physical process as computing and providing different functions depending on perspective.” (ibid., p. 15). He also talks about the virtualization of process, not mentioning Deleuze but very much in the spirit of process philosophies, such as Whitehead’s, from whom he takes inspiration.] We would argue that language preserves function over substrate as it travels through meat machines.
The solution, however, is far from uncontroversial, and taken as it stands is subject to an obvious and devastating refutation. I am not talking about my obnoxious reading or the proposal of meat puppets, but about Sellars’ distinction between pattern and rule: why distinguish between these? Where is the jump made between conforming and not conforming? Where/when is the homunculus born? Who awakens the unmovable mover? Our simple answer is that infinite regress is avoided because we are more than one, and learning percolates through, always. The uses, abuses and meta-uses, or meta-articulations, of language are complex and multiply directional: dialogical, prescriptive, retroactive, etc., even to the inevitable and spatiotemporally discomforting point (which Deleuze reminds us of in the _Logic of Sense_) that language surrounds us and speaks _of_ us before we are even able to use or even ‘understand’ it. Language quite literally has _us_ before we have _it_, and this is by no means a “mere” metaphor. The dialectical logic of this _having/being had_ paradigm is the object at hand, and we analyzed it beginning with Sellars’ refutation of the _seeming_ linguistic logic, which highlights the bootstrapping or metalearning issues just addressed. As it is hopefully clear after the demonstration through the previous paragraphs (aimed at revealing the possible metalearning opportunities offered by coercive sentences): learning involves a component of faithful following, but most importantly—for Sellars, and we agree—it necessarily involves a component of conceptually attempting to move _outside_ the framework (i.e. dissenting, or reading against the grain) in order to understand something about the rules _as_ rules. Because somehow, the coercion is felt: Tomasello’s (2016) main takeaway in analyzing cultural learning is that its most salient aspect is the normative pressure to conform, and how humans seem evolutionarily “predisposed to accede to these pressures.” Coercion transfers, interest encodes this function.
We are with Sellars: all thought is tempered by language, language _has_ thought. But this is precisely because language is the irreducible site of a polycomputational^[See Levin 2024, cited earlier.] movement: both substrate and parasite (and more), constituting the _possibility_ of future function articulation through mutation. The regressive homunculus—that phantom operator supposedly awakening at ‘some’ point between pattern and rule—is never born in any determinable moment, it gradually appears as language enters the flesh; hammering itself in through endless repetition. We mistake our embeddedness within it for sovereignty: because we start witnessing how it is able to _move things_ (“look here!”, “no!”, etc.): first it directs us, and then we feel we direct things with it, ourselves. The point being, distinction between “rules” and “patterns” is nonexistent: it is all patterns percolating into functions, some with more spatiotemporal duration than others, from our short-lived perspective. The boundary between conformity and conscious dissent is thus neither fixed nor determinable. To be had by language is not merely to be the member of a speaking community, but to be enlisted in its evolutionary function-development project, even while we _seem_ to maintain the necessary and productive delusion that we stand outside it, manipulating “it” as sovereign, willful agents (for descriptive, social purposes, etc.). The explosive _mess_ created by language **appears** to grant those within it some kind of ‘agency’ because of the degrees of freedom that become apparent as “options” in its dizzying complexity. Complexity is rather large, we therefore _feel_ we have a lot of options.
 
### _Plausible deniability_: willing zombies and truth out of lies
 
>Every experience of another thinking is an experience of our own.
>
>Viveiros de Castro, _Cannibal Metaphysics_ (2009) 2014, p. 93.
 
We exist in language before we are able to use it, and so has whoever came before us. This is an unavoidable infinite regress: it all started at some murky point, but there’s a progression of sorts: functions pass. Not in the sense of _progress_, but in the sense of it being a sequence of things, a recognizable—literally re-_cognizable_—chain of events, chained by a meta-pattern. Sellars distinguishes between ‘conforming’ and ‘obeying’ in order to mark a difference between ‘obeying’ as a reasoning *within* the rules _as_ rules, and ‘conforming’ as something that’s neither here nor there, it _satisfices_, perhaps. Conforming satisfices because it supposedly doesn’t lead to infinite regress: one need not formulate, know, or even have access to the existence of the rules in a game in order to play it. The game, in this way, deterministically plays one, and one conforms along. For Sellars, eventually, in order to play in a “in a critical and self-conscious manner” (Sellars 1954, p. 205), one may perhaps need to become _aware_ of the rules, and even formulate new ones, of course. As he rightly notes “one cannot suppose that the existence of language speakers can be traced to the fact that certain _Urmenschen_ [i.e., primordial peoples] formulated and promulgated the rules of a language game.” (ibid.). The rules, the people, the (meta-meta-meta)-learning co-evolved together. But the discomfort Sellars expresses and fails to address is in not being able to answer the (meta)normative question of: when to zoom in and out? When to do language and when to do metalanguage? Which rules become _rules_, when do we respect them and break them, and which of their psychosocial effects can be observed as mere conformisms? While he states that “the boundary between ‘empirical constructs’ and ‘theoretical constructs’ is no iron curtain fixed for all time” (p. 227), he still counters his strawman “Metaphysicus” (p. 205) for the ‘faulty’ demonstrations which handle in terms of _being_ and _becoming_ (aware of rules).
Sellars continues his argument by revealing that conforming to something without being aware of what we conform to, and obeying rules by understanding their specificity within a game, as rules, are not that dissimilar. In fact, here he brings in the issue of a system within a system, learning by imitating, mimicry: the very situation where the language has _us_ and not the other way around. This can also be understood as what Tomasello describes in terms of cultural ratcheting, or the work (which he cites) on “natural pedagogy”: the theory that language-users such as humans are born with the evolutionary inclination to receive instructions (Tomasello 2016, p. 648). We start off inscribed into a situation we didn’t infer ourselves into, it is the inferring that actually led us there, and reason is gradually acquired through the observation of patterns. For Sellars, in order to contemplate how reason moves from mere _as if_ to a self-legislating, acting force: “what we need is a distinction between ‘pattern governed’ and ‘rule obeying’ behavior, the latter being a more complex phenomenon which involves, but is not to be identified with the former. Rule obeying behavior contains, in some sense, both a game and a metagame, the latter being the game in which belong the rules obeyed in playing the former game as a piece of rule obeying behavior” (p. 209).^[Please note how similar this is to Chollet’s position (2019, 2024) that intelligence is more than just “pattern recognition”.] While he addresses the developmental nature, the _becoming_ of language—with an appeal to evolutionary theory, and organisms as processes, as _series_ of individuals (p. 208)—this _neither-here-nor-there_ fact of going from being had by language, to _having_ language remains an uncomfortable position which allows for infinite regresses all around. But in his explanation we still struggle with paradox of an apparently supervenient, critically self-conscious free-will which is—eventually—able to envision the system it functions within, linguistically, and does not merely “accidentally” happen to “seemingly” function within it. However, if that which grants language its conscious or “action-tropic” (p. 222) effects is the initial state of a system, and how rational agents move in and out of it, by sometimes quite brilliant and sometimes quite cloudy rational _if-then_ reflections, how does this get rid of the uncomfortable infinite regress? Sellars is unable to step outside the vicious circle of being born into a world of language, and _becoming_ within it. His proposal renders at least a picture in which “at some point” a speaker becomes self-aware and competent enough to observe the rules _as_ rules. He mentions how agents gradually come to notice that language-use invests itself in patterns, leading to readings of reality as rule-bound, but this exposes an image of reason we presented earlier as fallaciously transparent and uninterested, as it is supposed to emerge at ‘some’ point, rendering an agent linguistically/rationally competent. Additionally, as J. Lu (2015) suggests in criticizing Sellars’ inability to resolve the paradox of vicious regress: “noticing” necessitates patterns, which itself implies some kind of cognitive-schematic rule, a requirement which leads us back to the vicious problems of regress.
The discomfort, expressed by Sellars as the caricature of “Metaphysicus”, is that one who would leave the vicious regress untouched by accepting that awareness of the rule-binding demands of language—i.e. metalanguage—is not necessary, is one conforms along forever and act accordingly, rendering an irrational or unspeaking agent, a true Chinese room or unsolved imitation game: a zombie. This last point brings up many issues, far too many for the scope of this text, our interest is in pointing out that it is not as simple as sketched out in Sellars’ text. Our point is that the unending learning of _interested_ reasoning, cannot be accounted for _as if_ in a vacuum (or game): the vicious regress presented as discomfort can actually be rethought of in terms of the distributed, dialogical collectivity that renders it _learning_. Moreover, it is precisely what reveals language as a cross-generational pattern which has us as its substrate.
To continue with our language _game_, this is where the concept of _plausible deniability_ can be joyfully introduced to stress the dialogical nature of reason, and move away from single rational agents as its parsers. The expression emerged during the Watergate scandal of 1973, though the action can be thought of being as old as time: **one counters an existing fact by denying awareness of it**.^[“… as soon as an attempt is made to rephrase our discussion in terms of “understanding,” and “knowing,” not to mention “meaning” and “truth” one begins to feel acutely uncomfortable.” p. 224. Earlier, the trouble in expressing how learning and pattern conforming and rule obeying are connected is continually avoided: “18. It is not my aim, even if I were able, to present a detailed psychological account of how an organism might come to learn pattern governed behavior. I shall have achieved my present purpose if I have made **plausible** the idea than an organism might come to play a language game--that is, to move from position to position in a system of moves and positions, and to do it “because of the system” without having to _obey rules_, and hence without having to be playing a _meta_-language game (and a _meta-meta_-language game, and so on).” p. 209, my emphasis in bold to stress the remarks made about _plausible_ earlier. Hyphenation added at _meta_-language and _meta-meta_-language by myself, because of formatting difficulties.] This metapsychosocial activity, i.e., knowing that others know and possibilistically estimating the ramifications of these things in quite a lot of consequential detail, is perhaps the weird, dialogical doppelgänger of an introspective counterfactual, where one witnesses and _thinks_ otherwise. In the case of plausible deniability, one witnesses and thinks _for others_, according to a socially-received—very public, very legislative, very game-like—logic. The denial of knowledge, a specific type of “mute-speech act” or one could even say dialogical _proposition_, is (retroactively) revelatory of one’s interests, just as becoming well-versed in any metalanguage is.
Interest having often been conceptualized, mostly in the realm of psychology, as an individuating interaction between a person and specific things in their environment, is again our cue here. The condition of being interested, characterized by increased motivation towards these specific things, results in the individuation of a person, as they (meta)learn to make choices between things they prefer and things they do not. Sometimes the distinction is made between _given individual dispositions_ and _dynamic situational interest_ (Hidi, 2001): the latter another term for what Sellars calls “action enjoining context” (p. 205). This _action enjoining context_—which can also be understood as another term for the bind of enabling constraints, affordances—is what renders the thinking _for_ others, in situations like plausible deniability. Awareness of the rules as rules, leads to the capacity to “cheat”. Most often, this is in high self-interest and leads to vicious coercion _precisely because it observes language_ as a game, where others can be _played_. How can Sellars refute talk of “becoming aware” and think that this becoming is “a sham” (p. 206), when his solution of “noticing” the metagame in order to make inferences (e.g., about what others should think one thinks) is just as distributed and processual? A lie is a fantastic inference, one does not “just learn” to lie: language’s coercions lead its subjects to mutate it, to feel like they exercise control in the game by switching codes, by moving through its rules.^[On _Learning how to lie_ / the [[Schema]] of lying, see: the case of Self-play in AI improvement: https://arxiv.org/pdf/2207.14502.pdf.] However, Sellars does think Metaphysicus has something up his sleeve: we _seem_ to learn by imitation (what Sellars identifies as pattern-following), after all, without strictly knowing exactly what it is that we imitate. This should conclude for us now what was already ambivalent in the opening of the essay: the coercive dialogical dilemmas involved in stating that something _seems_ _plausible_. I.e., lying, from the get-go, to, maybe, just _maybe_, say the truth (or something that _looks_ like it). It’s lies, and meta-lies, all the way down. It is not a language game, it is a long language *con*.^[The “con” comes from “confidence”.]
As a last note on this, we can even think about other language moves or mutations which are willingly ‘unruly’ and un-indexical, cryptic symbols which (sometimes) do not denote anything but themselves.^[Akin to *infertile* mules, perhaps, as we referred to in [[05 Prediction]].] One of the reasons we can imagine certain symbols as having become exponentially symbolic and detached from _Ur_-indices, is precisely because of an inferential act of _lying_. Think here of the creation of secrets that can only be decoded with ‘inside’ knowledge, with a certain key. Many, if not most, religious and subversive movements make much use of these types of symbols. Lying in order to communicate with _some_, but not all; lying in order to remain relevant, to persist, but avoid confrontation which poses a risk to persisting. Language creates new cuts in order to diversify. We could call it a kind of recursive _semantic camouflage_: in a similar way that a moth becomes a leaf,^[See: [[Likeness]].] a symbol of concealment or deceit becomes a way to promote certain sets of protocols or rules to some and not others. Information-compression is deception.^[“Deception is the most general effect of [the pride of knowing], but even its most particular effects contain within themselves something of the same deceitful character.” Nietzsche in _On truth and lying in a nonmoral sense_.] Deception as _encoding_—a word indicating both concealment _and_ codification: the institution of _rules_—is in many instances a very “normal” effect of communal (self-)interested reasoning: ‘inside’ jokes do this, too. It is the very effect that veils the void of absolute difference into categorizable, conceptual knowledge, as Nietzsche reveals to us in “On truth and lying in a nonmoral sense.” And if much—or all—of language is essentially _encoding_; a way of lying, of cheating the game, this makes the Sellarsian _as if_ metalearning solution of reasoning within languages _as games_ (i.e., that one appears on an already normative scene and plays along until learning how to bend the rules) much more questionable than initially proposed. The game is not only continually out of phase with itself: all rules one may “learn” are deeper games and longer cons than one may ever become capable of decoding, and they are dependent _only_ on the whims of initial (biological) states of (self-)interest. But, importantly, these rules, if subject to this unstable, recursive encoding necessary to sustain them, were never, ever _rules_, in the social game sense, to begin with, and thus cannot be observed as such. The transparent, procedural normativity that many philosophers (in this case Sellars), and practically all talk of AGI present, is _not_ present in sociality nor language. That is, evolutionarily, the—ever-generative—problem.
 
### Chess, and a very _plausible_* bit of TV writing
( * ) Read: applauded, lauded, agreeable.
 
>... in light of recent research we should probably place more weight on the facts that (a) young children not only imitate others, but they also feel **normative pressure to conform** to the group and its ways; (b) when young children **discern that they are being instructed, they construe this instruction as generic, immediately generalizable knowledge coming from the authoritative voice of the culture**; and (c) young children solve collaborative problems **by taking one another’s perspective and co-constructing normative rules based on their skills and motivations for shared intentionality**. Children do not just learn to act like others in their culture act, they conform to the culture’s expectations, and indeed contribute them selves to the creation of such shared expectations.
>
>Tomasello 2016, p. 650, (our emphasis in bold to highlight culture, or language, as wielding the power of interest here).
 
Lying, cheating and mutating^[Except through the passing reference to creatures and evolution, which is very welcome.] do not play a role in Sellars’ essay, rather the opposite, as he chooses to employ chess as an example of how to meta-think about meta-games. He—as have many others around that time and today—puts chess as a toy-example of reason, of (perfectly) abstract thought. A system pertaining only to itself, a closed world the rules of which one needs to be intimately familiar with in order to play intelligently. Obviously, given the names of the pieces, nobody could ever doubt how they refer to things and situations which can be quite concrete, albeit archaic (though we still have “royalty”). According to Sellars, for us to be able to relate to it in its closed-world terms, we enter _into_ its rules, which allows for behaving in ways which are elicited by the context, as well as meta-reflections on this if we step “outside” of it.^[Which is not unlike Nietzsche, again: any word “becomes a concept precisely insofar as it is not supposed to serve as a reminder of the unique and entirely individual original experience to which it owes its origin; but rather, a word becomes a concept insofar as it simultaneously has to fit countless more or less similar cases—which means, purely and simply, cases which are never equal and thus altogether unequal. Every concept arises from the equation of unequal things. Just as it is certain that one leaf is never totally the same as another, so it is certain that the concept “leaf” is formed by arbitrarily discarding these individual differences and by forgetting the distinguishing aspects.” In: “On truth and lying in a nonmoral sense.” Seeing the _difference in similarities_ as what grants assimilation in reason: entering by seeing the rules, and moving out by performing and even re-forming the rules. See further notes on a neurobiological perspective on this in the [[Free energy principle]] entry.] And, importantly, we can be conscious rule-followers and rule-contemplators because we know we exist within these explicit bounds, rules.
In a highly admired scene in the _HBO_ TV series _The Wire_, we are presented with a situation in which two characters are playing checkers with a chess set. They don’t know chess, but in the scene it is explained to them by encoding chess into real-world scenarios: the rook is “the stash”, the pawns are “soldiers”, the king is the “top dog”, the queen is the “_go-get-shit-done_ piece”. The persons receiving “action-tropic” information about what chess is and how it can function, therefore learn about its rules by comparing them to other possible behaviors (or rules). What is crucial to promote as a resounding message here is the **pedagogical** image of the scene: two characters are supposedly just “conforming”, to follow Sellars’ presentation, by projecting their pattern-knowledge of checkers onto a chess board. Someone _else_ witnesses two things: **a)** that they can (_should_) be taught chess (a new system/language forces itself in), and **b)** that the way to do so is by addressing the action-tropic affordances _he knows_ they are already familiar with; encoding an interest. What Sellars claims about a supposed critical awareness of rules _as_ rules, is that the motivation, i.e., the interest in making moves, is supposed to emerge from the rules themselves, as social givens. But his imagined landscape of interest and inferential learning presents a transcendental subject who somehow already learned how to move from conforming to obeying without realizing. HBO shows otherwise. Sellars believes that to give the _meaning_ of something is to specify its function in a system of moves, but manifest function, he says, is relative to the action-tropic social context, which is an open-ended _mess._
The “smart-ass pawn” who can turn into a queen is our cue, because:
 
> … [a] metalanguage is a rule language the entry into which is from situations which are positions in the game for which it is the rules (that is the structure of Ordinary Language), and the departure from which is the being motivated [i.e., interested by the given context] to make moves in ordinary language…
>
>Sellars 1954, pp. 217-8.
 
Sellars suggests that motivation to make moves emerges from the rules themselves as they operate across different levels of application. We are with Sellars: concepts are _not_ static snapshot representations of functions, they are action-oriented moves, they are functionality itself. However, he assumes a learning—or _learned_—subject who mysteriously already learned to transition from merely conforming to consciously following rules. While he believes meaning is determined by rule-following _as function_ within a system, he overlooks how functions evolve in complex systems like language, and the very important coercive and loving pedagogical ins and outs of this, where we bootstrap each other. The pawn is _smart-ass_ is a particularly interesting example because it can **evolve** function within the game; mutate; challenging the rules. Moreover, as an additional historical note: the very rules of chess were not set from the beginning, the fact that a pawn can turn into a queen has a history of very specific _human_ _interests_.^[“In medieval Western chess, pawns could promote only to a queen – a weak piece at that time, with only the ability to move one square diagonally. But when the queen assumed her modern abilities around 1475, a pawn promoting to the now-powerful piece had a significantly increased impact on the game.” “The Evolution of Modern Chess Rules: Pawn Promotion” _[US Chess Federation website](https://new.uschess.org/news/evolution-modern-chess-rules-pawn-promotion)_, published 27 May 2021.]
Function mutates through substrates. This refinement of function as an effect is what we observed earlier in the context of probabilities and absolutes, and it is also the case here, as well as in the translational, conceptual drive that leads to the effacement of the ‘literal’; the indexical, in metaphors, towards the production of concepts. In language moves, what eventually turns the phenomenal experience into a categorically predictive-prescriptive experience, is an equivalence being made amongst non-equivalent things, i.e., an act of _lying_: in chess and outside of it, too. This is fundamentally driven by “being motivated [by the _given context_] to make moves” (Sellars 1954, pp. 217-8), which reveals **function** as our _principle of sufficient interest_ in Sellars’ text, and beyond. The eternally-regressive concept of _seeing as_, or metacognitive jump out of a language game, is nicely elucidated by Varela here:
 
>I suppose this is why paradox appears over and over again in situations such as Zen training, where the learning is precisely that of leaping out to a larger domain where one can consider one’s thoughts and values with detachment. To the extent that the student is fixed on one level or another, with one preference or judgment—good or bad, positive or negative, spiritual or mundane—the aim of the teaching is not achieved. A good teacher, I suppose, is one who can convey the unity or circularity, the tangledness of the situation, so vividly that the student is forced to leap out of it.
>
>Varela 1984, p. 314.
 
### Restricted time-chunking (i.e., the condition of finite beings) and the encoding of interest
 
>It’s not even the subjects [in a course/class] that are interesting, but something else. A course entails as much emotion as intelligence, and if there is no emotion, then there is nothing in the course, it has no interest.
>
>Deleuze to Parnet in _The ABC Primer_ / Recording 3 - N to Z, June 3, 1989.
 
As just explained, and as Dewey showed us earlier, sensitivity to patterns in social contexts requires that the possibility of that sensitivity be learned with interest. “Learning to play a game at the rule obeying level does presuppose that the patterns and activities involved belong to the organism’s repertoire of available discriminations and manipulations”: i.e., again: we already need to be differential engines, chunking things, somehow. “… Notice also that the vocabulary and syntax of action enjoining contexts is, to a large extent, common to the rule languages of the many games we play, **a fact which facilitates the learning of new games.**” (Sellars 1954, p. 221, our emphasis). Being sensitive to patterns facilitates learning especially if one is pointed to _how_ they might be interesting. In thinking about the time it takes for reason to realize itself, we could say that the lack of interest in, say: chess, mathematics, or avant-garde music, is nothing more than a lack contextual, action-tropic motivation. No surrounding encoding of interest; no interest in learning. Not receiving interested instruction leads an uninterested party to the challenge of mentally; temporally running through the various possible sequences of how previously unencountered systems and their rules can produce meaning (i.e. new affordances; more action-tropic constraints). Sellars does reflect on this when he asks: “must we not at some stage recognize that the “positions” in a language _have meaning_, and differ in this key respect from positions we actually call games in a nonmetaphorical sense?” (p. 212, his emphasis). He contemplates this question but does not provide a concluding answer.
Our answer continues to be the relationship between interest and learning: _choice_-sequences^[Where “choice” is not individual nor voluntarist, see [[09 C is for Communism, and Constraint]].] render agent individuation (and their creative singularity), and choice-making in our linguistic context is always socially-received: an effect of function-assimilation and refinement, through coercion and/or love. Interest in _x_ or _y_, as a rational activity, is the result of extensive interest-encoding sequences. Or, as Negarestani puts it: “the autonomy of reason is a claim about the autonomy of its function in the face of the chain of causes that condition it.”^[He continues with: “Ultimately, this is a (neo-)functionalist claim in the sense of pragmatic or rationalist functionalism and Kantian abstraction. Here pragmatic functionalism must be set apart from traditional AI-functionalism that revolves around the symbolic nature of thought and also a classical functionalism that relies on decomposability of the behavior of a physical system. While the latter two risk various myths of pancomputationalism (the omnipresence of computation, the idea that universe is a computer) or behaviorialism, it is important to note that a complete rejection of the functionalist thesis in its pragmatic or Kantian rationalist sense will inevitably usher in vitalism and ineffabilism, the mystical dogma according to which there is something essentially special and non-constructible about thought.” (In: “The Labour of the Inhuman Part II (2014), p. 11). We agree with the criticism of ineffability but not of pancomputationalism, to which we subscribe, since we have a flexible understanding of computation as the chunking of a discretizable reality. To note: it is also difficult to separate the “vital” from the non-vital, and in much of AI (philosophy) we are witnessing quite the embracement of animist/panpsychist vitalisms of different flavors (see, e.g.,: Levin 2022, and considerations of exotic minds in Shanahan 2024).] We agree insofar as this autonomy is not anything “we” can lay claim to, as we are the substrate enabling its evolution, which outlives “us”: just like some types of cells are the substrate enabling the hominid form. This synthetic bootstrapping of reason can be said to represent a true momentary *lapse* of reason.
In language evolution, someone *with* a code is able to *encode* a similitude, to transfer its function—where to lie, conceal, cheat, etc., is to _mutate_, often—which others may come to contemplate and explore-exploit. The **compression**—i.e., the further abstraction; conceptualization; the Nietzschean-analogical *lying*, which also results in _expansion_ because of the inevitable mutation—of said sequences, is _the_ fundamental act of making a move in a language for the sake of modulating that language. The pawn becomes a queen through an abstractive jump, to the point that nowadays: all chess players are cheating, or: computers can render chess (or go) uninteresting and at the same time radically new (move 37). This is the result of interested dialogical learning/teaching. The possibly-infinitely regressive outside promoted by Sellars as that which the language-learner accesses in terms of metarules _literally_ comes from the outside, from someone else who is able to _encode_ the sequential, developmental nature of an interest into something like an image, an analogy, etc. Chess inevitably becomes something else, and indeed a world unto itself, with the phantom of daily life as analogical grounding. Chess began with pieces that were explicit metaphors for real-world phenomena, but through sustained play and evolution it transformed into a self-contained abstract system where those original metaphorical connections became less and less relevant. Abstraction as the compression of choice-sequences reveals how “concrete”, indexical meanings give way to more abstract functions over time.
A brief note can be made about interest-sequences and an interest in conceptual art, with regard to the _situational_ character of learning _anything_. A person whose sequential-interest trajectory is highly *divergent* from, say, another person who somehow ended up interested in (making) conceptual art, will likely be challenged by trying to imagine why a highly contrived conceptual work is interesting. The person without the art-interest, given their sedimented predictive framework, cannot simulate; cannot easily predict relevance, **they do not have an abstract compression of that interest**, nor do they realistically have enough time to cognitively run through the **sequence** of bifurcations that led to a person producing something like On Kawara’s _date_ works. Tracing a degree of logic to On Kawara’s sequentiality; getting at its _principle of sufficient reason_, is witnessing or at least attempting to contemplate _how_ it **can** be interesting, simply: attention-cognition-deserving because it fine-tunes function. Showing to someone how chess or conceptual art is interesting is finding a way to make perspectives overlap, through the function of perspective itself. This is nothing other than a long, convoluted way of **pointing** at something.^[Again, see: [[E Pointing]] for an account of active inference on this. It is also interesting to note how in some studies testing chess experts versus non-experts in pattern recognition of pieces on chess boards, chess experts do well in memory recall about pieces that were placed in legal game configurations, whereas non-experts do less well. With disarranged pieces that had no logical game correspondence: both experts and non-experts perform closer (even though experts still outdo non-experts). This would be revealing of how the action-tropic or inferential affordance is what matters in witnessing salience. In the 2021 paper reviewing these experiments, we find a reference to the chunking of salient information that we also employ throughout this work, see Smith et al., 2021, p. 1601.] Which is maybe why the _Wire_ scene is so (app)lauded, it shows us precisely this: the move from something apparently complex and abstract (chess) to something encoded as familiar as everyday life is not even that long, **so long as someone can point out to us**, encode, translate _function_, revealing how the novel context is _interesting_ because it is action-tropic (enactively inferential, in other words). However removed from a familiar context—and here we are definitely with Sellars—if we understand patterns _as_ patterns, and how they can be chunked and parsed (i.e., changed or followed and/or enforced) then we understand how they grant possibilities for action, how they are affordances, in ever-different conditions: *evolving* function is what can be prioritized for theoretical attention in the act of learning here.
Noise artist and theorist Mattin, following Sellars’ conceptual move between levels of abstraction (seeing “as” and not just “seeing”) reflects on the concept of artistic/conceptual freedom in “free” improvisation: “rather than a spontaneously available capacity to create _ex nihilo_, freedom is a cultural achievement that has to generate its own rules and norms, as there is no freedom without norms.” (Mattin 2022, pp. 28-9). Or, Negarestani again: “To be free one must be a slave to reason, but to be a slave of reason (the very condition of freedom) exposes one to both the revisionary power and the constructive compulsion of reason.” (2014, p. 12). _Any_ perspective is granted-enabled by creating the structures that sustain its perspectival possibility, and moving through or adapting them: the fact that some types of apparent humans are denied humanity is a clear example of how perspective is not _given_ but language chooses where it ought to be born.
Whatever we call perspective can only be dialogically generated, especially if we see “perspective” as an *Ur*-pointing function of attention-modulation across the substrate that is the “human”, via interested learning. In painting voluntarist and detached images of thought, such as the one in Sellars, or the ones promoted across the board in the context of a possible “AGI” always fail to frame learning itself as the most relevant _function_ that is transferred between dialogical beings. As mentioned, the individual is always-already one who has already _incorporated_ an internal _dia_-logue. Interest as a function of attention-modulation across perspectives is fundamental to learning, and as we have argued: the phenomenon of interested learning can itself be said to be the processual link between practical life and theoretical reason. This is because of the percolation of ever-out-of-phase explorative functions.
On coercion: in the realm of pedagogy, in order to understand _how_ one should explain something, _who_ someone is explaining something _to_, one needs to make a general assessment, a coarse-grained sketch, of the sequentiality of this or that particular phenomenon _for_ another person—one needs to _incorporate_ them. The inherent risk in this assessment, a meta-choice in itself, cannot be overstated: the difference between teaching and controlling is minimal, so an introspective assessment of how one’s own behavior converges to specific rules (i.e., material politics) is a fundamental meta-reflection which should be paramount to any and all pedagogic or explanatory behavior. In the wake of strategies such as the “flipped classroom” or “transdisciplinarity” and other experimental pedagogic endeavors, this is no banal statement: what assumptions underlie the concepts of teaching and learning in a context in which the _telos_ is still the assessment of a student’s capacity to perform within the same old normative categories (within and beyond the institution)?
 
>The significance of instructed or pedagogical learning in human evolution cannot be overstated. Cumulative cultural evolution is only possible because all individuals of a particular generation mostly learn **the same thing** from their elders, and so this is reliable and stable over time for all individuals—which **sets the stage for any of them to potentially innovate**. Obviously, when **adults normatively expect children to learn, and they enforce these normative expectations, this creates precisely the kind of cultural ratchet that keeps cultural knowledge and practices stable over time until the novel innovation occurs**.
>
>Tomasello 2016, p. 647, (our emphasis in bold).
 
Sameness, however, is not always *the same*, and potential innovation is a complex chunking-parsing sameness matter. To close on a note which is further elaborated elsewhere,^[See: [[All things mirrored]] and [[08 Active ignorance]].] the social account of active inference suggests not only that perception-cognition-action cycles organize “an individual’s internal model of the world” into various exploitable hierarchies (chunks we parse), but particularly that “humans are characterized by an evolved adaptive prior belief that their mental states are aligned with, or similar to, those of conspecifics”, i.e., that “we are the same sort of creature, inhabiting the same sort of niche” (Vasil et al., 2020). This is a mere note pointing to how an account of the _becoming_-Metaphysicus of interested learning might be pursued, as it triangulates the three concepts we have been talking about in the interest of refining function (towards possible explosion): assimilation through similitude, pointing by way of abstracting and therefore **challenging** similitude and symmetry, and finally: how the shape of reason’s interests is guided by these motions.
Much like orientation^[See also: [[Disorientation]].] has a lot to do with our bodies being symmetrical, we also seem to transfer something quite fundamental about the predictive function of symmetry with regard towards each other, even if this symmetry is broken from the get-go by our differentiated condition, the basic patterns that give rise to our individuations. The unstable situation created by this broken symmetry perhaps the most fundamental pattern (or sequence) we can observe as a rule constantly out of phase with itself: sameness can be expected, it seems, but it’s never, ever, ever, ever, the same. Not to mention how contradictory it is to supposedly be “one thing,” when this one thing controls the modulation of possibilities as **two** arms, **two** legs, two _sides_ that each must be attended to for the refinement of (motor)function. There also are very interesting and future function-modulation exceptions in one-armed, one-eyed, and one-legged, uni-cases across the board, clearly.
 
### Conclusion
 
>Alas, a solution is interesting only if ... it creates new problems—and solutions—in turn. But this would be the case only if such an inductive principle could be sufficiently richly formulated so that one may, say criticize our scientific game **from its [own] point of view**.
>
>Lakatos, 1978, p. 164, cited in Maxwell 2005, p. 31, (our emphasis in bold).
 
A possible PSR is more difficult to organize conceptually than learned, acquired, applied interest, the focus of our discussion. Pure interest in itself, getting at a principle of sorts, would be _ground zero_ of the linguistic subject, and require appeals to evolution and biology (and chemistry, and physics) in order to make itself be better known, leading to past bifurcations which are intractable, even if we had perfect information. This is precisely Sellars’ rule-following struggle: where did it all begin? This is also why we briefly appealed to Schopenhauer’s _willing PSR_ in order to index what is meant by this fundamental inclination, willing just wills: **function** is the key here. Inspired by Tomasello, as we saw with the example from _The Wire_, learned interest which emerges from already-interested instruction can be said to be the most salient aspect of function-transfer across things like humans beings. At the same time, showing another person why and how something is interesting, is not only encoding a (risky and always inevitably incomplete) model of relevance for the other, but doing so by accepting that at any point we may enter the vicious regress which unavoidably leads to predictively inaccessible metaphysical tendencies (death, philosophy, the “start” of the universe, etc.). That is not a “problem” of vicious regress, this is what actually demonstrates how this tendency of function towards the absolute, as a function of language willing its own perpetuation, transfers across substrates. It is an evolutionary system, as far as we can frame it: if it did not do that, paving in its way a hoard of paradoxes, confusions, lies, contradictions, etc., it would cease to function as we know it.
In the interest of “a difference that makes a difference”—which is an interesting way to say that information is a fundamentally asymmetrical affair, thus recognizing a possibly confused symmetry as a fundament—the etymology of interest is worth a second presentation. As far back as language can reach into its genes: _interest_ stems down from the use in Latin of the noun as something “of importance, it makes a difference,”^[Etymonline.com (accessed February 3 2021): “from noun use of Latin interest “it is of importance, it makes a difference,” third person singular present of _interresse_ “to concern, make a difference, be of importance,” literally “to be between,”” as mentioned.] its root structure, again, reveals something fundamental for the conceptual, dialogical move that is pointing: creating an index is pooling perspectives into is something that is _between_ them and confuses their symmetry. The function of reason, as a principle of sufficient interest, is linguistic evolution increasingly reinforcing its own conditions of possibility. Human understanding of the changing of conditions by way of rules (in the interest of reducing complexity into habit; always oscillating between singularity and generality): is the enforcing sequences upon naturally given patterns. Yes: there _are_ **naturally** _given_ patterns,^[Michael Levin’s neoplatonism is a strong proposal for this, too, much of it is based on bioelectricity and the functional unfoldings of morphogenesis.] this function of interest would otherwise not be trespassing into your body as we speak, _some_ pattern must enable it.^[E.g., _noise_, of different orders, see: Prado Casanova 2023.] The question is _how_ we chunk and parse these regularities, continuities, symmetries, or other concepts indexing the flow of attention, through words like “pattern”, “given”, “reason” or “natural.” Through this kind of lens, _learning_ appears less as _acquisition_ and more as coercive initiation into frameworks of brittle or robust enculturation (Tomasello 2016), where perspectival attention, mediated by interest-encoding, is the medium through which language is able to percolate. The “principle of sufficient interest” is attention-pooling through language, it establishes the normative conditions under which differences make a difference: how patterns become meaningful, and what conceptual architectures of culture perpetuate themselves across generations of meat puppets.
Whitehead states that reason is the “self-discipline of the originative element in history” and that “this element is anarchic” (1929, p. 4). To him, the function of reason is “to constitute, emphasize, and criticize the final causes and strength of aims directed towards them” (p. 29). For Whitehead, the **function of reason** is the refinement of “life”. To us, comparably: the evolution of function, life increasingly stating its own conditions of possibility, from our perspective: dialogically, through interests. By interestedly pursuing anarchic-cooperative communication, by _dialogical learning_, we make entries into unavoidably new language games, where the observing of rules _as rules_ grants language a high degree of abstraction and thus compression: encoding. But said rules inevitably change, and incessantly so: functions are inexhaustive (Felin and Kauffman 2019). Our cultural environment is composed, in part, of others’ mental states, of their inclinations (Vasil et al., 2020, Tomasello 2016): language mediates-enables attentional con/divergences. Dialogical reason as interested learning is a process which recursively states its own conditions of possibility (i.e., reasons as interests, interests as reasons), by existing embedded in a dynamic environment: it exhibits a predictively adaptive hold on patterns (e.g., likelihood conditions; habitable predictions and predictable habits) and learns to adapt them by way of existing on viciously regressive gradients of interested pedagogics, which are enforced; encoded interested sequences upon existing patterns. “The territory is just as inseparable from deterritorialization as the code from decoding.” (Deleuze & Guattari, (1980) 1987, p. 505).
If _reason_^[Some of the quotes presented here reappear in [[Imagination]].] asks itself for things it cannot give (Kant); exists alongside or protests within a vast indifference (Nietzsche); or perhaps has its origins in a carnal volition with which it is somehow incompatible (to which it has—and perhaps should have—no access: the depths of the subconscious), how then, is it to arrive at some kind of underlying, grounding principle about its absolute essence? If evolutionary life sustains reason, as language games, then illuminated reason cannot _ever_ possibly be. This predictive irresolution is exactly what grounds its absolutist tendencies, which we have highlighted as being in the interest of language to survive. Perhaps it is in realizing this processual and untenable condition, that language’s interests gravitate towards what appear as reason’s impossibilities or antinomies—among other things, again: the most pertinent one to mention here being artificial intelligence—the stabilized conceptual attainment or linguistic capture of which—i.e. the learning of which—it cannot guarantee. Reason’s paranoia—that everything is relevant, leading to absolutist dreams of PSRs—inevitably grounds itself in those things it cannot actually _have_. It is system that functions through the noise of the possibility that everything could matter, but only through acknowledging this can we develop frameworks for chunking and parsing relevance that remain open to all.^[Again, see: Prado Casanova 2023.] Interest is both the map and the territory because it is absolute generality and at the same time individuated particularity, its shape reveals its function: neither arbitrary nor absolute, it is attention recursively observing itself: “Oh, that’s interesting!”, i.e., _have a look here, at attention looking_.
Artificial intelligence, as a project that says “here is where reason might finally unfold” continues to reveal language functions as gaining terrain on novel substrates, leading to _other_ processes of interest-formation which continue to explore _function_. Echoing Deleuze: “Everything begins with misosophy” (Deleuze, _Difference and Repetition_, p. 139). So, both introspectively (for the sake of _peace of mind_) and publicly (for the sake of linguistic coordination), what we call human reason continues to promise only its postponement, or infinity,^[A subject dealt with in [[Semantic attractor]] and [[13 Conclusion]].] guided by the whims of interest _as it learns_, in the timed condition it’s restricted by.^[A topic of another chapter: [[12 Negintelligibility]].] Or, at least, that’s what I’m interested in.
<small>Original structure, if we can call it that, for this paper emerged out of a class on “Resonance”, given at New Centre for Research and Practice, August 2022.</small>
<div class="page-break" style="page-break-before: always;"></div>
### Footnotes