Preterition

1. Exigency: a disagreeable subject

A language model produces a sentence. Ask the language model what it meant by the sentence and it will tell you. Ask again and it will tell you something else. Neither answer is a report, because there is nothing on which to report: no moment of meaning, no interior from which a sentence may emerge, nothing hidden behind the text to be revealed. Rhetoricians have responded to this absence almost entirely with questions about provenance: who or what made this, and how does that making count?

A question about provenance is a question about a subject. It assumes that an embodied intention fixed the meaning at the moment of production, that the interior of that intention exists, and that a text can be read back toward that interior. By operating as the writer whose interior we cannot pretend to read, the language model has made visible a habit that was never as safe as it looked: assuming a specific interiority is and always has been unreliable. If reading a text back to that interior was always unreliable, what were we trying to achieve when we did it?

What now would we need to assume instead?

For my purposes here, I begin with what a theory must assume before it can be applied. Any account of economy needs a subject: someone whose wanting, thinking, choosing, and doing the account explains. Any account of writing requires, as well, a theory of a meaning and intending subject. Across economic and writerly theories of a subject, however, the accounts that argue most loudly about what defines the subject say much less about what they permit themselves to assume they can know about that subject

The ideologies of mainstream neoclassical economics rest on revealed preference: the act discloses the interior, completely, at the instant of choosing. Thorstein Veblen mocked that subject in 1898 as “a lightning calculator of pleasures and pains” and “a homogeneous globule of desire.” The critique still stands. Veblen aimed at interiority’s presumed contents and the notion that the subject’s interiority is a hedonic ledger. But the contents are not the load-bearing parts: behavioral economics has since replaced them by swapping the calculator for a bundle of biases and heuristics. More recent economic theories of the subject still read the interior from the act, and some behavioral economists now claim to read interiority better than the subject. Veblen’s dig partly missed, because it aimed at the furniture and not at the floor.

I argue we should name the floor as an access regime: what a theory lets itself assume it can know about an agent, from what evidence, and at what moment. Two theories can share an access regime and disagree about everything else. The revealed preferences of neoclassical economics and the all-consuming nature of Knapp and Michaels’s intentionalism in “Against Theory” have almost nothing in common—one is about choosing, the other about meaning; one puts preferences inside the agent, the other inside intentions—yet both hold that the relevant interior state is fully and instantly available from the act that expressed it. Two theories can also agree about the contents and split on access. Keynes and his neoclassical contemporaries both grant that agents hold expectations about the future. Keynes says nobody can know them, the agent included, because about such matters we simply do not know. Opponents of Keynes say the expectations are there, stable, and readable. Same furniture, flipped floor.

Once we see the floor, the pattern in the first case becomes nameable as presumed synchrony: the assumption that a subject operating over time can be read at a moment in time, and that the moment counts as evidence. The assumption of revealed preferences and the assumption that makes authorial intentionalism possible rely on the same synchrony. A theory of writing as an economy cannot borrow a subject from either assumption, because every subject on offer arrives with a pre-installed access regime: an OEM protocol for interiority. A theory of writing requires a subject specified by its access regime rather than by its contents: a subject materially particular, formed in time that subject cannot recover, whose interior is real and whose accessibility is too variable—from any position, including the subject’s own—to permit stable construction of that interior by a reader. The term I have chosen is the embodied agnostic subject.

In today’s arguments around the LLM production of language, the machine is the occasion, not the reason. The reason is that people operate over time; nothing that operates over time can be read at a moment without being falsified in the reading. The writing that follows seeks specification over application: I set out here the minimum subject upon which a responsible account of writing and its value can rest. Every account of meaning that reads a text as a window into a writer builds on a substrate that resists inspection. The accounts that do so most confidently conclude that uninspectable writing carries no value.

2. Necessary: what the subject must do

Before this theory of a subject can work, there are three conditions that must hold true of that subject: it has a body, it was formed in time it cannot recover, and it exists in a structure of address.

Judith Butler’s “Giving an Account of Oneself” supplies all three. It answers, in advance, the objection that if the self is opaque to itself, it cannot be held responsible. Butler replies that the opacity is what produces ethical responsibility. The subject is opaque to itself by virtue of its relations to others, and those relations are the venue of its ethical life rather than an obstacle (2001, 22).

The opacity is first temporal: the norms by which the subject becomes recognizable are not born with that subject. A subject’s story begins in the middle and arrives late. A subject reconstructs that subject’s own emergence and cannot be present to it, and tells the story differently each time, and no single telling is the one for which the subject is accountable (2001, 26-27). In the book that followed the article, Butler puts the general case plainly: the “I” trying to account for itself finds a self already implicated in “a social temporality that exceeds its own capacities for narration” (2005, 8). The embodied agnostic subject is historical not because it has a past, which is trivially true of all things, but because the subject was made in time it cannot recover.

Disability rhetoric gives the body a great deal of content—a neurology, a sensorium, particular processing tendencies—and shows that access demands fall on bodies unevenly and expensively (Price 2011; Dolmage 2017; Yergeau 2018). Cavarero’s exposure, which Butler borrows for singularity, gives the body no content at all beyond the irreducible fact of being this one, here and now, in public (Butler 2001, 25). Theories of an opaque subject carry a seam between content and no content.

History closes that seam from both sides at once. The body’s content is real, and it is deposited as what a history left behind in a particular entity. Edouard Glissant supplies the figure: what settles at the bottom of the mirror is “a whole alluvium deposited by populations, silt that is fertile but, in actual fact, indistinct and unexplored even today” (1997, 111). Deposit is both fertile and, to the ground that carries it, unexplored. Particularity is what history leaves in a body, and non-negotiable variance is that same history’s irrecoverability to its deposited body.

An implied accompanying claim: the body was present. It is the only thing that was there from its own formation, through every continuous moment of the formation. It is therefore the best-positioned witness to its formation. It simply cannot testify. Butler’s late-arriving narrator is not someone who missed the event, but someone who was at the event for the duration and cannot return. The party with maximal proximity to the evidence is the party every institutional demand for interiority addresses. The demand trades on the proximity and ignores the irrecoverability.

The third precondition is address. An account is always given to someone, and the giving dispossesses it: sent and made recognizable in terms that are not one’s own (Butler 2001, 26). Butler is precise about the effects of these constraints on the genre: giving an account “takes a narrative form,” draws on “narrative voice and authority,” and is “directed toward an audience with the aim of persuasion,” and the narrative does not report a prior moral agency but is “the prerequisite condition” for having any to report (2005, 12). An account of oneself is a rhetorical performance. It was never a report.

The embodied agnostic subject, then, must at a minimum carry three capabilities. First, hold and acknowledge the body’s particularity without converting that particularity into a claim about the body’s contents that the body itself could not make. Second, hold variance as the subject’s own temporality, so that a report at one moment cannot stand in for the subject at another. Third, as Butler suggests (2001, 37), the subject must refuse to let responsibility depend on transparency.

Further implications of the embodied agnostic subject will follow in a subsequent post. I engage some of those implications in a blank verse dialogue elsewhere on this site.

References

Butler, Judith. 2001. “Giving an Account of Oneself.” Diacritics 31 (4): 22–40.

Butler, Judith. 2005. Giving an Account of Oneself. New York: Fordham University Press.

Cavarero, Adriana. 2000. Relating Narratives: Storytelling and Selfhood. Translated by Paul A. Kottman. London: Routledge.

Dolmage, Jay Timothy. 2017. Academic Ableism: Disability and Higher Education. Ann Arbor: University of Michigan Press.

Glissant, Edouard. 1997. Poetics of Relation. Translated by Betsy Wing. Ann Arbor: University of Michigan Press.

Keynes, John Maynard. 1937. “The General Theory of Employment.” Quarterly Journal of Economics 51 (2): 209–23.

Knapp, Steven, and Walter Benn Michaels. 1982. “Against Theory.” Critical Inquiry 8 (4): 723–42.

Price, Margaret. 2011. Mad at School: Rhetorics of Mental Disability and Academic Life. Ann Arbor: University of Michigan Press.

Quiggin, John. 2010. Zombie Economics: How Dead Ideas Still Walk among Us. Princeton, NJ: Princeton University Press.

Veblen, Thorstein. 1898. “Why Is Economics Not an Evolutionary Science?” Quarterly Journal of Economics 12 (4): 373–97.

Yergeau, M. Remi. 2018. Authoring Autism: On Rhetoric and Neurological Queerness. Durham, NC: Duke University Press.