Five Readers, One Chamber: The Predicting Mind in Dream of the Red Chamber

In January 1927, in Xiamen, Lu Xun wrote a short preface, Jiangdong Huazhu Xiaoyin, for a stage adaptation of Cao Xueqin’s unfinished novel, and in it summed up a century of arguments over the book in one sentence: 经学家看见《易》，道学家看见淫，才子看见缠绵，革命家看见排满，流言家看见宫闱秘事. The classicists see the Book of Changes, the moralists see indecency, the romantics see tender longing, the revolutionaries see anti-Manchu sentiment, the gossips see secrets of the palace. Five readers, one novel, five books with almost nothing in common.

The easy response is relativism: a book means whatever its reader brings, so naturally a moralist and a revolutionary leave with different novels. That response skips the interesting question. The question is not whether readers diverge but how one fixed sequence of characters yields five coherent and mutually exclusive books, rather than one book misread four times. Philosophy of mind has spent several decades building a fairly precise answer, and it begins with the observation that reading never happened on the page alone.

The first layer is what it is like to read the words at all. Two readers looking at the closing line of the novel’s prophetic song, 落了片白茫茫大地真干净, “and all that is left is a vast white emptiness, truly clean,” are not having the same experience because they are looking at the same characters. What philosophers call qualia, the felt quality of an experience, is precisely what a shared text cannot hand over. A reader who has watched a family’s fortune collapse brings a weight to “clean” that a reader who has known only comfort cannot produce on request. The Mary’s Room thought experiment makes the same point about color that applies here to grief: no description, however complete, substitutes for having lived what it describes. The text gives every reader the same symbols. It cannot give them the same felt reading, because that was never in the text. It belongs to the nervous system doing the reading, and no two of those are the same instrument.

That gap would guarantee some variation, but it does not explain why the variation settles into five stable camps instead of scattering into noise. A second idea is needed. Edmund Husserl, and later John Searle, treated intentionality, the fact that consciousness is always consciousness of something, as basic to mental life. The literary theorist Wolfgang Iser applied the idea to reading. A text, he argued, is not a complete message but a structure full of deliberate gaps, and the reader’s directed attention fills them. Cao Xueqin left an unusual number open. The reader is never told with any precision what Lin Daiyu looks like beyond a scattering of poetic gestures, and never given access to what Miaoyu thinks behind her fastidiousness. A mind directed at such gaps fills them, automatically and mostly unawares, with whatever material that particular mind already has.

Cognitive science now has a specific account of that material and how it is used. Predictive processing models of the brain describe perception as inference rather than recording. The brain holds prior expectations about how the world is usually arranged, predicts its incoming signals against them, and revises only when the mismatch, the prediction error, becomes too large to absorb. Iser’s gap and the predictive brain’s unresolved prediction are close to the same object, described once by literary theory and once by neuroscience.

The mechanism becomes concrete when particular priors meet particular passages. A classicist steeped in the Book of Changes and its correlative cosmology reads the second chapter’s careful account of the Jia household’s ranks and branches, or the symmetries of the twelve beauties of Jinling, and expects order of a cosmic kind. The narrator supplies no gloss on these enumerations, and the prose offers just enough parallelism that the expectation is never contradicted. Nothing in the text produces the surprise that would force the classicist to revise.

A late-Qing revolutionary reads the same pages differently. Where the classicist sees cosmic cycles, a reader formed by anti-Manchu pamphlets and anxiety about the dynasty’s collapse sees a coded political allegory in the decay of a great house, its corrupt ceremonies and its fall. Cao Xueqin includes real official titles, Manchu naming customs and a long institutional decline, and attaches no explicit moral commentary, so nothing in the text contradicts a political reading either. Neither reader is careless. Each is doing what a predicting mind does, on a text whose underdetermination is unusually well judged.

This is also where the extended mind thesis, the argument by Andy Clark and David Chalmers that thinking can run through objects outside the head, earns a place without any exotic claims. Dream of the Red Chamber is a book people return to over decades, and the familiar experience of rereading it, when a passage that meant nothing at nineteen reorganizes a whole chapter at forty, is a loop. The book’s open gaps stay where they are while the reader’s priors change with life, and each return to the text changes the priors again for the next reading. Rather than mirroring a fixed reader, the book works as an external store that shifts the terms of the reader’s future inferences.

That still leaves the clustering: why the readings settle into a handful of camps. Consider the scene in which Xue Baochai urges Jia Baoyu toward an examination career and he rebuffs her. Readers whose model of other people is tuned to social obligation find Baochai’s advice sensible and Baoyu’s refusal childish. Readers whose own minds run closer to Baoyu’s find the reverse. The capacity to model other minds by simulating them with one’s own runs most smoothly on characters who resemble minds the reader already knows how to simulate. Lu Xun’s five camps are five stable points where a particular set of priors, a particular style of social simulation and the novel’s deliberate omissions lock together without producing enough friction to force revision.

Taken together, this changes what the five readers are evidence of. They are not evidence that literary meaning is relative, that any reading is as good as another because the text constrains nothing. A carelessly made book resolves such disputes quickly: supply enough disambiguating detail, prediction error rises, and readers converge. What is remarkable about Dream of the Red Chamber is how long it has avoided that, across two centuries and millions of readers, by holding open enough gaps in character, symbol and motive that the classicist, the moralist, the romantic, the revolutionary and the gossip can each read to the end without meeting the correction that would bring them together. That is ambiguity precise enough to house five incompatible readings at once and let every one of them close the book convinced it read correctly.

Lu Xun counted five. He could not count the reader of this sentence, who brings priors of their own about consciousness, inference and gaps, and fills Cao Xueqin’s omissions with material he never imagined. That reading is no less real than the classicist’s or the revolutionary’s. It is only the one running now, in a mind that is, like the other five, quite sure it is reading the book correctly.

Editor’s note: the same prior-driven mechanism this essay traces in five readers of one novel has a direct experimental version. Adventurous or Reckless? primes you with one of two word lists before asking you to judge an identical, deliberately ambiguous stranger.