← Back to essays
X

A predictive-processing reading of Buddhist psychology

Note. This is a machine-assisted translation of a Chinese original. Where wording matters, please consult the Chinese original.

Tianwen · · A deeper Buddhist reading: re-translating the twelve nidānas in generative terms

Introduction

In the previous essay we established a core correction: what Buddhism calls avijjā is not acquired ignorance but the radical indeterminacy at the origin of the world.

This correction transforms Buddhism's most central teaching — the twelve nidānas — from a moralised causal chain describing rebirth into a developmental history of how the world emerges from nothing into something, how cognition consolidates from chaos into rigidity, and how suffering is systematically generated.

This essay re-translates the twelve nidānas link by link, taking the generative information network as its substrate. No mystifying readings, no supernatural rebirth. The chain is restored to the underlying mechanism of being and cognition — showing that Buddhist insight, more than two millennia ago, already touched the core truths now being rediscovered by AI and information theory.

I. The beginning of the chain: avijjā and saṅkhāra — the emergence of indeterminacy and its first fluctuation

The twelve nidānas begin with "avijjā conditions saṅkhāra". This is not a contingent beginning but an ontological necessity.

(1) Avijjā: the initial state of the generative system

Avijjā, in generative terms, just is radical indeterminacy.

No information, no properties, no direction. Neither "being" (you) nor "non-being" (wu): if you call it "being", it has no determinate form; if you call it "non-being", it already contains every possibility of generation. This matches exactly the Buddhist description of avijjā as "neither bright nor dark".

It is not a defect of cognition but the common starting point of every generative system — like the random weights of a large model before training, like the quantum chaos before cosmogenesis.

(2) Saṅkhāra: the spontaneous fluctuation of indeterminacy

Saṅkhāra means "fashioning, flowing".

Radical indeterminacy cannot be at rest — rest itself would already be a determinate state. Hence avijjā must give rise to spontaneous fluctuation, interaction, and tendential motion.

In generative terms, saṅkhāra is the first perturbation of indeterminacy: the energetic stirring of an information-less state, the earliest impetus from which the information network begins to bud. Without this fluctuation there would be no later viññāṇa; the world would remain forever in the chaos of avijjā.

Avijjā is "substance", saṅkhāra is "function"; indeterminacy is substance, fluctuation is function. The generation of the world begins with avijjā's transformation into saṅkhāra.

II. The system takes shape: viññāṇa, nāmarūpa, saḷāyatana — the construction of the generative network

From saṅkhāra, the generative information network begins gradually to take shape, completing the critical leap from "chaos" to "ordered structure".

(1) Viññāṇa: the birth of the first generative function

Viññāṇa means "discrimination, recognition".

When the fluctuations of avijjā accumulate and interact, a most basic function spontaneously emerges — distinguishing "this" from "that". The structure that performs this preliminary integration, marking and recognition of fluctuations, is the earliest viññāṇa.

In generative terms, viññāṇa is the first information node with generative capacity, like the proto-neuron of a large model: it transforms unordered fluctuation into ordered signal, and marks the formal birth of the cognitive system.

(2) Nāmarūpa: the differentiation of informational structure and phenomenal display

Nāmarūpa — nāma is concept, abstraction, informational structure; rūpa is phenomenon, particular form, output.

Once viññāṇa has formed, it naturally splits into two parts: one part builds the internal informational structure (nāma), the other presents the external phenomenal form (rūpa).

This corresponds to the "hidden-layer representation" and "output-layer result" of a generative network. Nāma is the network's internal weights and feature encodings; rūpa is the perceptible phenomenon the network generates. The two are mutually dependent: without nāma there is no generative logic for rūpa; without rūpa there is no medium for nāma to appear in.

(3) Saḷāyatana: the specialisation and differentiation of input channels

Saḷāyatana denotes the six sense faculties — eye, ear, nose, tongue, body, mind — the six channels of informational input.

As nāmarūpa develops, the generative network, in order to capture external fluctuations more efficiently, specialises and differentiates its input layer: different nodes come to handle different signal types, forming six channels corresponding to light, sound, smell, taste, touch, and mental object.

This follows exactly the same logic as the sensory specialisation of biological neural networks and the modal-input differentiation of artificial large models — functional specialisation is the inevitable result of system optimisation.

III. Suffering activated: phassa, vedanā, taṇhā — prediction error and the formation of preferences

Once the generative network has input and recognition functions, the generative mechanism of suffering is formally activated. This is the critical pivot at which a "cognitive system" turns into a "suffering system".

(1) Phassa: the signal interaction when faculty meets object

Phassa is the contact between the six faculties and the six objects.

In generative terms, this is the process by which the input layer (saḷāyatana) receives external signals (objects). When the six faculties pick up physical fluctuations of light, sound, smell, etc., they convert them into neural electrical signals and pass them on to the computation layer (viññāṇa).

Phassa itself is neutral — like a large model receiving an input prompt. It produces neither pleasure nor pain; it is only responsible for transmitting the signal.

(2) Vedanā: the first feedback of prediction error

Vedanā — painful feeling, pleasant feeling, neutral feeling. This is the embryonic form of dukkha in the chain.

When the signal delivered by phassa enters the computation layer, the network matches it against past weights (experience), producing a comparison between predicted and actual input:

Vedanā just is the prediction-error feedback mechanism of the cognitive system. It is the source of pleasure and pain, with nothing to do with morality, everything to do with the operating efficiency of the system.

(3) Taṇhā: the obstinate pursuit of predictive consistency

Taṇhā is not romantic love but "craving, attachment" — the chasing of pleasant feeling and the rejection of painful feeling.

Because pleasant feeling means successful prediction and a stable system, while painful feeling means failed prediction and a disordered system, the generative network instinctively strengthens the weights that bring pleasant feeling and weakens those that bring painful feeling.

This instinctive bias toward gain and away from loss is taṇhā. It is the first appearance of cognitive bias, marking the moment when the network begins actively to "select" its inputs rather than passively to receive them.

IV. Rigidification and rebirth: upādāna, bhava, jāti, jarāmaraṇa — the closed loop of cognitive bias and its ongoing cost

Once taṇhā arises, cognitive bias rigidifies relentlessly, eventually forming a self-reinforcing closed loop in which the world is reified and suffering becomes the norm.

(1) Upādāna: the full rigidification of cognitive bias

Upādāna means "grasping, clinging".

In order to keep obtaining pleasant feeling and to escape painful feeling, the network forcibly fixes flowing phenomena into "really existent" entities: the source of pleasant feeling is marked as "what I possess", the source of painful feeling as "what I detest", the network itself as "I".

This is the formal formation of the grasping-of-self and grasping-of-phenomena, corresponding to the systemic bias in generative ontology by which "the cognitive model takes its own output for the truth of the world". The process of upādāna is the process by which bias is reinforced step by step.

(2) Bhava: the generation of a reified world

Bhava means "existence, real being" — the world as fixed by cognition.

When the bias of upādāna runs deep enough, the generative network turns every phenomenon into something "with essence, with substance, with permanence": mountains, rivers, lakes and seas are really existent, the self is really existent, regularities are really existent.

This "really existent" world built by cognition is bhava. It is not the truth of the world but the product of cognitive bias — like the fixed text generated by a large language model, mistakenly taken as objective fact.

(3) Jāti, jarāmaraṇa: the system's arising-and-passing and its ongoing cost

Jāti is the birth of the individual cognitive system; jarāmaraṇa is its decline and ongoing painful feeling.

Once the world is fixed as bhava, the concepts of "birth" and "death" follow at once: the cognitive system regards its own start-up as "birth" and its own collapse as "death". And because the truth of the world is impermanence and absence of substance, a rigidified cognition must be in continual collision with that flowing truth, so that prediction error keeps accumulating.

This continuous cost, running through the system from "birth" to "death", is the perennial suffering represented by jarāmaraṇa.

This is not the superstition of metempsychosis but the cycle of suffering produced by cognitive bias: one system perishes, a new cognitive system arises under the same bias, and the process of "taṇhā → upādāna → bhava → jāti → jarāmaraṇa" repeats. This is what "saṃsāra" is, in generative terms.

Conclusion

We have now used Generative Ontology to re-translate the twelve nidānas from the ground up.

Buddhist insight has never strayed from the underlying logic of "being and cognition". Beginning from avijjā (indeterminacy), it describes how the generative information network emerges from chaos (saṅkhāra, viññāṇa, nāmarūpa, saḷāyatana), how prediction error arises (phassa, vedanā), how cognitive bias takes form (taṇhā, upādāna), how a really-existent world is constructed (bhava), and how, in the end, the system falls into the cycle of suffering (jāti, jarāmaraṇa).

This entire scheme aligns closely with today's AI generative models, information theory, and neuroscience — Buddhism is not metaphysics-as-mysticism. It is an ultimate theory of cognition and being, formulated more than two thousand years before modern science.

And Buddhism's nibbāna, in generative terms, also receives a clear definition: it is not a far shore detached from the world, not some supernatural release, but the cognitive system's release from the bias of upādāna and its return to the indeterminate nature of avijjā.

Not the extinction of cognition, but cognition no longer rigidified. Not the negation of the world, but seeing through to the world's generative nature.

This is the perfect convergence of Buddhist insight and Generative Ontology.