Where does one moment end?

Imagine slowing Leela’s response until every change is visible. At first, you may expect the true boundaries to reveal themselves. That expectation hides a choice: what must change before we count a new event?

The invisible cut

A camera can divide one second into 24, 60 or 1,000 frames. The road does not acquire more events when the frame rate rises. Yet some events do disappear at low resolution. A brief flash may fall between two exposures. A rapid hand correction may blur into one movement. The camera therefore discovers some temporal structure and creates some of it.

A cognitive taxonomy faces the same problem. We might draw a boundary when the object changes, when a new function begins, when causal influence switches direction, when report becomes possible, or when a tradition says one citta has ceased and another arisen. Those criteria need not agree. Temporal granularity is part of the theory, not a neutral window onto it.

The word moment can therefore carry at least five commitments. It may mean only that one event precedes another. It may add that later events depend on earlier ones. It may mark a functional change, such as orientation giving way to discrimination. It may claim a measurable duration. It may finally claim that reality itself is composed of indivisible temporal atoms. Sliding between these meanings is easy because the same diagram can represent all five.

The ladder supplies the paper’s central discipline. When a source describes citta after citta, we first ask whether it asserts order, conditional passage, functional role, metric duration or ultimate ontology. We then mark what kind of evidence supports each rung. A traditional manual can authoritatively state its own sequence. Historical research can locate that sequence within a textual development. Neither source measures neural timing.

This move is conservative rather than reductive. It does not say that the Theravāda tradition intended only a modern functional model. It says that different parts of the inherited account have different evidential routes. We can preserve the tradition’s own stronger claims as tradition-internal commitments while refusing to smuggle them into neuroscience or engineering.

flowchart LR
  A["The red lamp that never became<br/>a thought"] --> B["Where does one moment end?"]
  B --> C["What the sequence actually claims"]
  C --> D["Turn the map into an intervention"]
  D -. "re-test" .-> A
  classDef first fill:#0a1424,stroke:#087A98,stroke-width:2px,color:#e6f3ff
  classDef step fill:#0a1424,stroke:#3f7da0,stroke-width:1px,color:#e6f3ff
  class A first
  class B,C,D step

What the sequence actually claims

The familiar citta-vīthi diagram is not a transcription from a high-speed instrument. It is a highly organised model from the developed Theravāda Abhidhamma tradition, codified in commentaries and manuals from materials and categories with earlier roots.

That status matters. The canonical Abhidhamma provides extensive classifications of consciousness and conditional relations. The detailed process courses found in later Theravāda expositions are set out in commentarial literature and compendia, including the medieval Abhidhammattha-saṅgaha. Bhikkhu Bodhi’s modern manual explicitly explains that its cognitive-process chapter relies largely on the commentarial tradition. Contemporary scholarship likewise describes the mature five-door and mind-door sequences as later systematisation rather than a simple quotation from the early discourses. Broader studies by Noa Ronkin and Y. Karunadasa supply historical and systematic context for that development.[1][2][9][10]

Build the course from the red lamp

Return to Leela. Before the lamp flashes, her conscious life is not absent. The Abhidhamma model uses bhavaṅga to describe a passive mode of consciousness that occurs between active cognitive processes and in dreamless sleep. Rupert Gethin warns that translating it as “unconscious” is misleading. It is not a blank, not a hidden warehouse and not a modern subconscious. It remains intentional consciousness with an object within the traditional account.[3]

The new visual object then disturbs that passive current. The standard account of a very clear visual object lays out seventeen positions. The first three concern bhavaṅga. Fourteen process cittas proper follow, beginning with five-door advertence. The course eventually returns to bhavaṅga. These are logical positions in the model, not milliseconds measured from Leela’s nervous system.

The sequence is easiest to understand as a set of transitions. The lamp first affects the passive flow. Five-door advertence turns the course towards the relevant sensory door. Eye-consciousness sees the visible object in the technical sense supplied by the model. Receiving takes up what has appeared. Investigating and determining differentiate the object’s significance. Javana is the active phase, ethically and kammically charged in the traditional system. Registration briefly continues with the same object before the flow subsides.

These glosses are not interchangeable with common psychological words. “Investigating” is not necessarily explicit reflection. “Determining” is not a little person issuing a verdict. “Registration” is not automatically long-term memory. Javana is especially easy to flatten into a modern “decision module”, but its place in Theravāda thought includes moral quality and karmic efficacy. Paper 107 can examine the associated mental factors; this paper needs only the temporal role.

Not every object completes the course

The process becomes more causally interesting when the object weakens. In the manual tradition, “very great”, “great”, “slight” and “very slight” do not name physical size. They concern the force or clarity with which the object impinges and how far the process can run. A very clear object reaches registration. A somewhat weaker one may end after javana. A slighter one may reach determining but not the active javana phase. A very slight object may merely disturb bhavaṅga without producing a full cognitive process.[1]

This is the sequence’s most useful causal feature. If every input always passed through every stage, the taxonomy could be a fixed narration laid over behaviour. Conditional endings instead imply gates. They invite a question: can we produce a change at an early stage while selectively preventing later consequences?

Continuity without an owner

The process model also addresses continuity. A sequence of conditioned events can form a lawful course without a permanent controller standing behind it. The manual’s traditional mango simile describes a sleeper awakened by a falling fruit, then looking, taking, examining, eating and returning to sleep. The point is not that a tiny self performs each step. The steps are connected by conditions. The agent-like unity belongs to the organised succession, not to an unchanging inner owner.[1]

This is consistent with the wider Buddhist rejection of a permanent self, but the detailed momentary account is historically developed. Modern scholarship distinguishes early teachings on impermanence and dependent continuity from later, more radical doctrines of momentariness in Buddhist scholastic traditions.[2][4] The distinction prevents a common shortcut: citing an early statement that all conditioned things change, then treating a later seventeen-position sequence as if the same statement supplied it in full.

Turn the map into an intervention

A historical taxonomy becomes practically useful when it tells us which alteration should change which outcome. Without that step, almost any continuous process can be redescribed as a procession of named moments.

The Temporal Process Schema

The proposed artefact has two layers. The first records the inherited claim without translation. The second states a candidate functional mapping and its test. This separation matters because a failed modern mapping should not be reported as a refutation of the historical tradition, and a successful engineering decomposition should not be reported as proof of its metaphysics.

Each schema instance names a candidate boundary, trigger, ordered stages, optional exits, observables, interventions, rival generators and permitted conclusions. It also assigns every assertion to one rung of the Temporal Claim Ladder. A comparison is admissible only when its claim level is explicit.

type ClaimLevel = "order" | "gate" | "role" | "metric" | "ontology";

type Stage = {
  id: string;
  traditionalLabel?: string;
  functionalGloss: string;
  entersWhen: string[];
  exitsTo: string[];
  observables: string[];
  interventions: string[];
};

type TemporalProcessSchema = {
  candidateBoundary: string;
  trigger: string;
  stages: Stage[];
  claimLevels: ClaimLevel[];
  matchedRivals: string[];
  allowedInference: string;
  prohibitedInference: string[];
};

The schema encodes one invariant: an observation cannot silently climb the ladder. If an experiment finds a reliable stage order, the result is stored as order. Fixed duration remains unset. If ablation of an intermediate function selectively blocks later effects, role gains support. Ontology still remains unset.

A small causal assay

Consider a synthetic alert process with four possible endings. Salience below 0.25 only perturbs baseline. From 0.25 to 0.49 the process reaches determination. From 0.50 to 0.79 it reaches active response. At 0.80 or above it also produces a retention trace. These thresholds are deliberately invented. They expose the logic of a gated model without pretending to reproduce Abhidhamma measurements.

Now intervene on the determining stage. If the active response disappears while earlier detection remains, the stage distinction earns some causal support. If every intervention simply crashes the entire system, the test is uninformative. If a simpler two-stage rival predicts the same effects, the fine-grained sequence has not yet earned its extra complexity.

What the lab tests

The lab tests three propositions. First, an ordered process can remain stable while stage durations vary. Second, object clarity can change the terminal state without changing the stage vocabulary. Third, an interruption can distinguish a staged route from a bypass that reaches the same output. These are claims about a synthetic model, not about ancient chronology or human physiology.

A positive result means the proposed ordering and gate are useful for explaining this configured system. A negative result has several meanings. The stage may be causally idle. The intervention may be too broad. The observable may not distinguish the rivals. Or the process may be better modelled as continuous dynamics without sharp intermediate states. A failed assay lowers confidence in the mapping, not in every interpretation of the source tradition.

A realistic configured-system example

Suppose a railway crossing monitor receives a camera flash that may be a signal lamp, reflection or sensor fault. A practical system separates raw detection, source routing, evidence receipt, classification, policy determination, response proposal, authority check, typed action and effect receipt. This resembles an event grammar because later actions depend on earlier transformations and can stop at several gates.

The resemblance is useful only if its limits remain visible. The monitor’s authority kernel is external and deterministic. Its evidence receipt is an auditable record, not the Abhidhamma’s tadārammaṇa. Its proposal stage has no traditional karmic meaning. The comparison can still generate a test: remove classification while preserving flash detection. A lamp-specific closure command should fail, while a generic anomaly alert may remain.

stateDiagram-v2
  [*] --> Claim
  state "The invisible cut" as Claim
  state "Build the course from the red lamp" as Test
  state "Not every object completes the<br/>course" as Revision
  Claim --> Test: expose mechanism
  Test --> Decision: evidence survives
  Test --> Revision: rival survives
  Revision --> Claim: narrow or rebuild
  Decision --> [*]

Where the frames dissolve

The tempting modern move is to match citta moments with perceptual cycles, neural oscillations or computation steps. That move compresses several unsettled questions into one visual resemblance.

Rhythmic sampling is not a citta clock

Contemporary perception research contains serious proposals that perception or attention is periodically sampled. VanRullen reviews behavioural and neural evidence for perceptual cycles. Herzog and colleagues propose a two-stage model in which quasi-continuous unconscious processing is followed by slower discrete conscious percepts. Fekete and colleagues argue that evidence offered for discrete perception can remain compatible with continuous dynamics. Menétrey, Vogelsang and Herzog later emphasise that researchers must specify exactly which aspect of discreteness a brain-wave result is meant to support.[5][6][7][8]

This disagreement is not noise to average away. It reveals the same ladder problem. A periodic modulation of detection may support rhythmic sampling. It may not show that perceptual content updates only at those phases. Even discrete content updates would not show that consciousness itself consists of equal, indivisible frames. None of those findings identifies a traditional citta.

The metaphysical interpretation remains underdetermined as well. A physicalist can treat a successful stage model as a functional decomposition realised by continuous biological dynamics. A biological naturalist may require organism-specific processes before the model says anything about experience. A functionalist may care mainly about preserved causal roles. An idealist can regard the sequence as an articulation of changing appearances within awareness, while a process philosopher can treat events and relations as more basic than enduring substances. The same ordering evidence can sit inside each view because it does not decide what consciousness ultimately is. Temporal structure constrains an ontology only after a theory explains why that structure is constitutive rather than merely correlated.

This is also where a consciousness-primary orientation must remain disciplined. It may ask whether cittas are modifications, disclosures or delimitations within awareness rather than producers of awareness. That is a serious alternative research question. The citta-vīthi diagram alone cannot settle it, just as a neural rhythm alone cannot settle physicalism. The present contribution therefore stops at the event grammar and records the metaphysical residue rather than erasing it.

The strongest rival: post-hoc segmentation

Any sufficiently detailed output can be narrated as preparation, uptake, appraisal and response. A researcher who sees the result first can choose stage boundaries that make the sequence look prescient. The strongest rival explanation is therefore not “nothing happens”. It is that the citta-inspired segmentation is one of many equally adequate descriptions of a continuous or differently organised process.

The remedy is not more terminology. Pre-register the mapping, the intervention and the rival. State what would count as selective preservation of an earlier stage, what downstream effect should disappear, and which simpler decomposition could explain the same data. The sequence gains scientific value only when it excludes some possible courses.

An evidence matrix for responsible status

The same proposition can be legitimate under one evidence status and illegitimate under another. “The manual places determining before javana” is a textual claim. “A modern classifier has an intervention-sensitive gate before action” is an engineering claim. “The classifier implements the same kind of citta” is a structural or ontological claim requiring much more.

What would strengthen or weaken the proposal?

The event-grammar interpretation would strengthen if several observations converged: stage order remained stable across tasks; weaker inputs reliably truncated later functions; selective interventions spared upstream observables while altering predicted downstream effects; and the same schema generalised across sensory doors without changing its rules after the fact.

It would weaken if stage boundaries moved freely with the analyst’s labels, if multiple simpler schemas predicted every intervention equally well, if supposed stages were only decodable but not causally used, or if evidence for rhythmic sampling was the sole bridge to discrete consciousness. A rival continuous-dynamics model may produce apparent stages when read through thresholds. That rival must be implemented, rather than only named.