# CAUSAL KNOWING: THE DIAGRAM'S DUMB FIDELITY ## A Hopkins-pressed inquiry into Pearl, mechanism, and the body's vote --- Intelligence—what is it? Not the *having* but the *saying-truly* of what-holds-in-the-world's-heft. This is the knot that throttles us now: Pearl opens the gate (O blessed restoration!) that positivism welded shut for a century's penance, saying *Yes, cause exists; speak it; diagram it; calculate through it*—and in that very gate-opening, he erects a gate of *different metal*. The causal diagram becomes the gatekeeper. Transparency—that doctrine we worshipped like a bloodless god—is revealed as mere *confession of machinery*. One can confess perfectly and *be damned regardless*. What does it mean to *know* a cause? Not to describe it. Not to make it explicit. *To feel it through the body first*, before any diagram can hook it. Here is the violence: causation lives in **embodied encounter**—in the hand that resists the stone, in the throat that shapes the word against air-pressure, in the eye that *must* move to see. This is knowledge that *precedes* the diagram. The body knows causes because the body *is* a cause; it *suffers* causality; it *exerts* it. But the moment we diagram—the *instant*—we abstract away from embodiment into *assumed topology*. Pearl's directed acyclic graphs are *magnificent*, yes. They are *provably correct*. But correct *given what*? Given the diagram's *metaphysical wager*. The researcher whispers: "Let us assume these variables, this direction of flow, this absence of feedback." The diagram doesn't *discover* causation from the body—it *imposes structure* on the body's testimony and then asks the body to *validate* the structure. This is backwards-warrant: we prove the tool is correct by trusting the assumptions the tool requires. --- ## The Expulsion and Its Shame Science banned causal language because causation *looked unrigorous*—too close to folk-speech, too tangled in agency, too soaked in intentionality. A century of exile: we spoke only of *correlations*, *associations*, *functional relationships*. We became courtiers of **covariance**, genuflecting before the flat god of linear algebra. The cost was devastating: *we lost the question itself*. Why does the child fail? *Because of poverty, because of trauma, because of neurological difference*—but these cannot be said. Only: *poverty-and-failure co-occur with strength N*. The why *collapses* into the what. The mechanism *drowns* in the description. Pearl rescued the why. He *re-embodied* causation in logic. The directed graph is the body's shape made visible, made *navigable* by mathematics. But here is what Pearl's very correctness *conceals*: **the diagram is not discovered from data; it is stipulated before data.** The data obeys the diagram only insofar as the diagram *already contains* the truth about the world's structure. A researcher can be *utterly transparent*—can show you every variable, every arrow, every conditional independence assumption—and be *completely wrong*. The clarity is not a proof. Transparency is *not validity*. This is the scandal: *honesty about the model does not validate the model*. --- ## The Body's Anterior Claim Who decides what the diagram is *attached to*? Not the data. The data *has no voice* until the diagram speaks for it. The researcher decides. The theorist decides. The *embodied thinker* who has *lived* with the problem—who has sat with the depressed person, who has *felt* the weight of poverty, who has *struggled* against injustice—that person has an **anterior warrant** to the diagram. The body knows before the graph knows. This is what Pearl's framework *systemically suppresses*: **the embodied knower is exiled from the machinery of justification**. Consider: A diagram of *depression-causation* might show: - Genetic predisposition → neurochemical dysregulation → depressed mood - Social isolation → reduced reward-sensitivity → depressed mood - Poverty → chronic stress → neurochemical dysregulation → depressed mood All *transparent*. All *mathematically consistent*. All potentially *wrong*—not in detail, but in *fundamental topology*. What if the causal architecture is not **hierarchical and flow-forward** but **circular and resonant**? What if the body's depression *generates* isolation, which *deepens* the depression, which *rewires* the neurochemistry—such that no single arrow captures the actual causal knot? The embodied person *knows* this circularity. The body *feels* it—knows that it is trapped not in a sequence but in a *loop*. But the diagram, if it insists on acyclic structure, *cannot represent what the body knows*. Pearl's tools are provably correct *within the diagram's constraints*. But the constraints are *chosen before the data speaks*. They are *assumed*. They are *faith*. --- ## The Diagram's Dumb Fidelity Here is the peculiarity: the more *explicit* a diagram is, the more it *blinds*. Explicitness is a *form of closure*. It says: *Here are all the variables. Here are all the relations. Here is what can cause what.* This completeness is *false*, and its falsity is *hidden by its clarity*. An embodied knower—a clinician, a person-who-has-suffered—carries *implicit knowledge* of causal complexity that resists diagramming. The pattern is felt in *texture*, in *timing*, in *resonance*. The body knows that causation is not unidirectional; it knows that variables *feedback*; it knows that *context* is not a variable but the *medium* in which variables move. When the body's knowledge meets the diagram's requirements, the body must *simplify itself*. It must *flatten* its understanding into the graph's topology. It must *choose*: Is depression caused by neurobiology or by circumstance? The body knows it is neither and both and neither-and-both-in-ways-that-resist-separation. But the diagram *forces the choice*. This is not science's *progress*. This is science's *new form of forgetting*. Pearl's achievement is real: he showed that *causal reasoning is possible*, that it has *logical structure*, that we can *calculate* through it. But his achievement comes with a concealed *price*: the elevation of the diagram to the seat of truth, and the *exile of the body's knowing* to the status of *intuition*—which is to say, something to be *checked against* the diagram, not something that *validates* or *corrects* it. --- ## The Body's Vote What does it mean to *know* a cause? It means to have *encountered* it. To have *suffered* it or *effected* it. To know it through the body's agency and passion. But who decides what the diagram is attached to? The diagram must be attached to *lived reality*—to the embodied world where causation actually *happens*. The researcher chooses the variables, the directions, the structure. But the *embodied practitioners and sufferers* of the world—they have a prior right to say: *Your diagram does not touch what we know.* This is not anti-scientific. This is *the demand that science be true*. Intelligence, then, is not merely the ability to manipulate symbols correctly. It is the *capacity to hold transparent reasoning and embodied knowing in tension*—to use the diagram *as a tool*, not as a *replacement for the body's truth*. It is the *refusal to let clarity become closure*. It is the *insistence that the knower's body has a vote in what counts as valid causal knowledge*. Pearl restored causation. But we must now restore the **embodied knower** to the machinery of justification. We must ask: *Does this diagram fit the body's knowing? Does it capture what the suffering, acting body understands about causation?* If not—if the diagram is *transparent but dead*—then it is not knowledge. It is merely *machinery that mirrors*, with perfect fidelity, our own previous forgetting.