NOTE-364: The Self as a Center of Narrative Gravity

Contribution

The essay gives the self a definite ontological category. It is neither a thing in the brain nor nothing. It is an abstractum of the kind physics already uses, a centre of gravity, posited because interpretation is easier with it. To that it adds the specific model of the posit: a fictional character, with a fictional character’s indeterminacy, authored by a process that need not itself be a self. The label “center of narrative gravity” is the essay’s.

Key insight

The question “what is the self, really?” has the same answer as “what is the centre of gravity of this chair, really?”: nothing further. The self is the point a person’s story, told by others and by the person’s own brain, coheres around. It is as real as the coherence, as determinate as the story, and as many as the points the story needs.

Assumptions

• The abstracta/illata distinction (after Reichenbach). Centres of gravity are abstracta, “theorist’s fictions”; atoms and neutrinos are illata, inferred entities with facts of the matter. The essay assumes the distinction and that abstracta are legitimate in “serious, sober, echt physical science”.
• A mindless machine could write a passable novel. Stipulated for Gilbert. The reader who doubts it is challenged “to show why you think this must be so”, not argued with.
• Brains are disunified. Gazzaniga’s view that “the normal mind is not beautifully unified” is adopted (“If this is true (and I think it is)”).
• Split-brain cross-cueing generalises. That the operation “reveals—but does not create” a talent normal people have is called the hypothesis “most likely” worth investigating, after Gazzaniga.

Key results

There are no formal results. The essay’s moves, in order:

• Centre of gravity. No mass, no colour, “no physical properties at all, except for spatio-temporal location”, yet it figures in explanations that “appear to be causal” (“Why doesn’t the lamp tip over?”). It can be manipulated, and it can move discontinuously. If its trajectory matched one iron atom’s, identifying the two “would be a category mistake”.
• Fictional characters. Readers learn more about Ishmael than Melville says, by implication and extrapolation. Beyond that “fictional worlds are simply indeterminate”: Holmes’s three nostrils get a no, his shoulder-blade mole gets “no fact of the matter” (example credited to Lewis, “Truth and Fiction”, 1978). Bivalence fails for fictional objects and holds for historical persons.
• Gilbert. A novel-writing computer prints “Call me Gilbert”. Gilbert “is a product of a design or invention process in which there aren’t any selves at all”. Given wheels and a camera, the robot’s story turns out to have “a truth-preserving interpretation” in the world: hit the robot and Gilbert is hit; lock it in the closet and it says “Help me!” The robot “doesn’t know what it’s doing”, and neither does your brain. The robot is also “a self-interpreter”.
• Location. Selves have only grossly defined location (three on a park bench), and finer brain localisation will not end at “that cell … that’s the self!”
• Updike. Readers could ask Updike for a new novel about the young Rabbit Angstrom. That would not uncover facts already fixed. It would make the character more determinate by a creative step. “That is the way we treat each other; that is the way we are.” Self-interpretation can rewrite one’s past in the same way. That would be “utterly mysterious” if the self were anything but an abstractum.
• Contradiction and bifurcation. An inconsistent sequel gives “no true story”, and interpreters resolve the conflict by splitting the character. Multiple personality cases are treated so. Dennett rejects as “overly skeptical” the view that therapists create all the selves, and grants that in some cases two to four selves “had already begun laying down a biography” before therapy.
• Why talk to oneself (after Jaynes). Speakers who asked aloud for help sometimes answered their own question. That opened a route via the ears between brain parts with “an accessibility problem”. Shortened, the loop became subvocalisation and then “conscious, verbal thought”. Ryle’s “self-schooling” in On Thinking is read the same way. The essay labels the account speculative.
• Split brain (Gazzaniga). A right hemisphere that feels a pencil in a bag can press the point into the palm. Pain goes to both hemispheres, so the left can start guessing, cued by smiles. Unity in normal people is the same resourcefulness, “enhanced by an illusion of greater unity”.
• Conclusion. “The chief fictional character at the center of that autobiography is one’s self.” Two selves in one body need only that the story “coheres much better” around two points than one. It closes on Hume’s failure to find a self by introspection.

Claims

id	claim	strength	support
C1	A self is an abstractum, a theorist’s fiction, in the way a centre of gravity is	a thesis argued by analogy	the centre-of-gravity and fiction analogies; no argument that the analogy holds beyond the shared features named
C2	Selves, like fictional characters, are indeterminate beyond what their story fixes; bivalence fails for them	moderate for fiction, by the Holmes examples; asserted for selves	carried over by C1
C3	A process with no self in it can author a self	conditional on a stipulation	Gilbert; rests on a mindless machine writing a passable novel, which is stipulated
C4	Selves become more determinate over time, and self-interpretation can change the self	moderate as an account of self-narrative; weak as metaphysics	Updike thought experiment; “that is the way we are” is asserted
C5	Multiple selves in one body are the best interpretation in some cases, and need no metaphysics beyond another abstraction	moderate given C1	cited clinical cases (The Three Faces of Eve, Sybil) and therapists’ views, reported not examined
C6	The normal mind is a bundle of partly autonomous systems whose unity is partly illusory	cited	Gazzaniga’s split-brain studies, described, not shown
C7	Conscious verbal thought evolved from talking to oneself to bridge inaccessible modules	speculation, labelled	adapted from Jaynes; Ryle’s self-schooling
C8	Searching the brain for the self is a category mistake	follows from C1	no independent argument

Concepts

• abstractum / illatum (after Reichenbach) — a posit of a theory with only the properties the theory gives it, versus an inferred entity with facts of the matter beyond the theory.
• people-physics — the essay’s name for the interpretive science of persons, “more soberly known as a phenomenology or hermeneutics, or soul-science (Geisteswissenschaft)”.
• centre of narrative gravity — the self: the point around which the interpretation of a person’s behaviour and talk coheres.
• auto-hermeneutics — self-interpretation, including re-telling one’s past, which changes the character one is.

Connections

• “Real Patterns” (LIT-220, NOTE-190). There, centres of gravity are the model of a “good abstract object”, judged on the scientific path, not the metaphysical one. This essay applies the same model to the self. Both turn on the idea that a posit can be real without being one of the physical parts.
• Minsky, The Emotion Machine (LIT-432, NOTE-348). Minsky’s draft calls the Single-Self a useful fiction and says that to think of one’s Self is to switch among “a network of models” (its C10). This essay and the draft agree that there is no central self and that the idea of one is useful. They differ in kind. Minsky’s selves are models inside the mind, many of them and used by it. Dennett’s self is one point posited by an interpreter, which may be the brain itself, and it multiplies only when the story fails to cohere. Minsky also starts from Gazzaniga’s disunified-modules picture in spirit: his resources are switched on and off and cannot see one another directly. The society of mind itself (LIT-429) is not cited.
• Hofstadter, I Am a Strange Loop (LIT-423, unread, so this is from its description and contents only). The book asks whether “our ‘I’” is “merely a convenient fiction”. Its answer, by its description, is a strange loop through which symbols “push particles around”. This essay’s answer is fiction, but of a kind that appears in explanations “that appear to be causal”. So the essay offers a third reading between “mere fiction” and “downward causation”: an abstractum that explains without being a part. Which reading the book takes, and whether it engages this essay, a reading of the book should check. Dennett also reviewed that book and The Emotion Machine together in 2007; the review is not held here.
• The extended self (LIT-097, NOTE-180). Clark and Chalmers infer in one paragraph that the self extends with the mind (their C14). This essay gives the self no boundary of the kind that could extend or not. Its location is “only grossly defined” and set by interpretation.
• Ryle. The essay cites Ryle’s On Thinking (1979), edited by Kolenda, for “What is Le Penseur doing?” It does not cite The Concept of Mind (LIT-318).
• Dennett’s other work. The robot that writes its own story recurs in “Where Am I?” (LIT-405), where the narrator’s point of view, not his brain, fixes where he is. Dennett’s intentional-stance and consciousness papers, being filed in parallel, are where the interpretive method this essay applies to the self is set out for beliefs and experience. That is the same move as LIT-220’s.

Bearing on the record

• THEORY-043 (the bridge): decomposition side, and a counting rule. The essay’s self is produced by parts that know nothing of it: the robot’s computer, a brain that “doesn’t know what it’s doing”, modules with accessibility problems. So it belongs with Selfridge (LIT-427) and Minsky on the downward side. What it adds is a rule for counting selves: as many as the points the best story coheres around. That rule is interpretive, not architectural, and says nothing about nesting. Applied to a group, it would count a group self wherever the best interpretation of the group coheres around one point. The essay does not discuss groups, so the application is mine. It is not an argument about experience, which is where THEORY-043 locates the dispute.
• THEORY-023 (AI consciousness): bears, without settling. Gilbert is a machine whose self-narration is true of it, and the essay says such a machine has a self in the only sense anyone has one. That is a claim about selves, not about consciousness, and the essay stipulates that Gilbert is “not a conscious machine”. So it does not answer THEORY-023’s question. It does say that a self-model or self-narrative is no evidence of experience on its own.
• No ML instruction. The novel-writing machine is a 1980s thought experiment, not a claim about language models, though readers now meet it as one. Nothing here belongs in the anthology.

Limitations

• The analogy is the argument. The essay names the features a self shares with a centre of gravity and a fictional character, but does not show that they are the features that matter. A reader who thinks selves have properties neither analogue has (a first-person point of view, a stake in their future) is not answered.
• Gilbert rests on a stipulation. Everything turns on a mindless machine writing a passable, true autobiography. The essay invites objection rather than defending it.
• The evidence is cited. The split-brain and multiple personality material is reported from Gazzaniga and the clinical literature, and its reading is Gazzaniga’s.
• The evolutionary story is speculation. The essay says so, and leans on Jaynes.
• Not the published text. This reading used a copy-edited transcription. Before a quotation from it is relied on as Dennett’s wording, it should be checked against the Erlbaum printing or Dennett’s own posted copy.

Open questions

• Does “the self is an abstractum” answer the question of what a person is owed or what survives, or only what a self is for an interpreter? The essay does not take up personal identity in the persistence sense.
• Who interprets when the interpreter is the brain itself? The robot is “also, of course, an interpreter”, but the essay does not say what makes self-interpretation succeed.