The bell is one event only from a distance

There is a harmless sense in which the four accounts concern the same event. The same body occupies the room. The same bell vibrates. The same student later reports hearing it. That shared reference matters, but it does not supply a shared decomposition. A meteorologist, a sailor and a painter can all point to one cloud while tracking pressure, navigational risk and reflected light. Their agreement about the cloud does not make pressure, danger and colour rival names for one hidden component.

The philosophical case is harder because words such as mind, awareness, consciousness, feeling and intention already sound like components. Translation encourages the reader to line them up. Yet Pātañjala Yoga groups cognition, affect and volition within the changing functions of citta, while Sāṅkhya places consciousness proper with puruṣa, not with those functions.¹² Early Buddhist analysis begins instead from conditioned events such as sense faculty, object, consciousness, contact and feeling, then asks how appropriation and suffering arise without a permanent owner.³⁴⁵

Chinese materials resist even the convenience of one “Chinese map”. In Xunzi, the senses differentiate features while xin, the heart-mind, knows, judges, intends and can direct the body. Correct names are conventional achievements constrained by discriminable patterns and sustained through education and ritual.⁶⁷ Zhuangzian texts place pressure on the confidence of any such scheme by exposing how distinctions and standards depend on situated perspectives.⁸ To write “the Chinese view” is therefore already to erase a live argument.

Husserlian phenomenology changes the task again. Its initial question is not which causal component produces the bell experience, but how the bell is given as this meaningful object. The sound is encountered against a horizon of what has just elapsed, what is anticipated and what else could be perceived. Intentionality names this directed structure. Retention and protention help explain why a temporal object is experienced as unfolding rather than as unrelated instants.⁹ Phenomenologists also distinguish prereflective self-givenness from turning the self into an object of reflection.¹⁰

Why the architecture metaphor arrives too early

In engineering, an architecture is expected to remain identifiable across several tasks. A cache does not become a scheduler because the operator’s goal changes. This stability makes architectural comparison powerful. It also makes the metaphor dangerous when imported into traditions whose categories are partly diagnostic and transformative. A category may identify what must be noticed, weakened or relinquished, not a permanent module waiting to be located.

Consider pain. A clinician may partition the episode into tissue damage, nociception, affect and report. A meditation manual may partition it according to sensation, aversion, identification and release. A phenomenologist may distinguish bodily location, affective colouring, temporal persistence and the narrowing of one’s world. These descriptions can share observations without sharing a component list. The clinical map is organised around diagnosis and treatment. The practice map may be organised around changing one’s relation to the episode. The descriptive map asks how pain appears before a causal theory is selected.

An architecture claim therefore carries an extra burden. It must show that the proposed units remain causally useful when the practical aim changes. If a boundary disappears as soon as the task changes, the boundary may belong to the method of inquiry rather than to a task-independent mechanism. Conversely, if the same intervention repeatedly separates the same relations across aims, a stronger architecture hypothesis becomes reasonable. Stability across purposes is evidence for architecture; resemblance within one purpose is not.

The consequence is not that comparison is impossible. It is that comparison needs a stronger unit than the translated noun. The four maps can still constrain one another where they make claims about the same controlled variation. Does removing lexical recognition alter the felt tone? Does interrupting temporal carry-over destroy the melody? Does attention training change the transition from feeling to impulse? Does restraint require an enduring witness, or can it arise from conditioned processes and cultivated dispositions? These questions cross vocabularies because they specify a change and an expected difference.

A shared architecture should be treated as a hypothesis earned by preserved interventions, not as the starting premise of translation. This moves the inquiry from “Which word means consciousness?” to “Which relation would have to remain stable for these two accounts to explain the same change?”

What survives when the nouns are removed

Suppose a translator makes a four-column glossary. Citta, viññāṇa, xin and intentional consciousness appear on one row under “mind”. The table is tidy and nearly useless. Citta can designate a changing psychomental complex whose operations are witnessed by puruṣa. Viññāṇa is dependently arisen and differentiated by sense domain. Xin can include cognition, evaluation, desire and command. Intentional consciousness is described through its directedness and modes of givenness. A single English noun has hidden four different boundary decisions.

Remove the nouns and write each proposal as a relation. For example: a differentiated sensory condition contributes to an episode; the episode acquires significance through retained context; affective or evaluative tone changes the next tendency; training can alter the transition from tendency to action. These claims remain imperfect, but they can now be challenged by changing one condition at a time.

A comparison record with break conditions

The smallest useful record has six fields. It names the trigger that brings a relation into play, the transformation claimed, the consequence expected, the practical purpose for which the distinction matters, the exclusion or error the map is designed to prevent, and the ontological commitment left by the account. This turns a loose analogy into a dossier that can fail.

Consider a simple pair. Record A says: “a sound and trained convention trigger recognition; recognition makes a summons available; the summons guides norm-sensitive restraint; the purpose is cultivated action; the account excludes raw sensation as sufficient; the heart-mind is the relevant integrator.” Record B says: “a retained auditory profile triggers fulfilment of an anticipated object; the tone appears as a meaningful summons; the purpose is description of givenness; the account excludes an instant without horizon as sufficient; ontology is bracketed at this stage.”

The pair shares a relation between temporal carry-over, learned significance and the appearance of a summons. It differs in purpose and in what counts as an explanation. The result is qualified correspondence, not agreement about a component called mind. The correspondence strengthens if manipulating learned convention changes anticipatory structure in both reconstructions. It weakens if one account predicts that the practical meaning can vanish while its decisive relation remains untouched.

This discipline resembles the best comparative work in philosophy: seek enough commensurability for disagreement to be intelligible, while resisting both descriptive chauvinism and the perennialist claim that all traditions secretly say the same thing.¹¹¹² A recent bridge between the Buddhist five aggregates and contemporary consciousness research offers a useful precedent. Its authors propose functional mappings that generate hypotheses while explicitly refusing to claim historical identity or metaphysical equivalence.¹³

The standard is therefore not perfect translation. It is purpose-preserving comparability: enough shared structure to design an intervention, enough recorded difference to interpret a negative result, and enough restraint to leave ontological residue visible.

flowchart LR
  A["The bell is one event only<br/>from a distance"] --> B["What survives when the nouns<br/>are removed"]
  B --> C["The seams are more informative<br/>than the overlap"]
  C --> D["Build comparisons that can fail"]
  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

The seams are more informative than the overlap

Once local correspondences are visible, comparison often becomes too polite. Writers celebrate process, relationality, embodiment or non-duality, then quietly step around direct opposition. The seam is where the inquiry becomes useful. Pātañjala Yoga depends on discriminating puruṣa, the seer, from the transformations of citta. Early Buddhist not-self teaching instructs the practitioner not to identify form, feeling, perception, formations or consciousness as self.⁵ These are not two poetic descriptions waiting for a better synonym. They differ over what liberation requires one to discern.

Husserlian prereflective self-awareness does not repair the opposition. It concerns the non-objectifying way experience is given as one’s own, not necessarily a permanent substance standing outside mental change.¹⁰ Treating it as a Western name for puruṣa imports precisely the ontology that phenomenological description may bracket. Xunzian xin differs again. It is an active centre of discrimination, evaluation, intention and bodily direction inside a programme of cultivation, not an inactive witness isolated from nature.⁶

Five outcomes, not one similarity score

The consistency matrix uses five verdicts. Agreement means the records preserve the same intervention-sensitive relation at a compatible grain and purpose. Qualified correspondence means a relation is shared but its grain, purpose or exclusion differs. Conflict means the accounts make opposed claims under a common question. Translation failure means no stable relation has been identified, often because one term spans several roles or the aims are incommensurate. Metaphysical remainder means operational and phenomenological relations can be aligned while an ontological claim remains undecided.

These verdicts are not a ladder from bad to good. Conflict can be the most informative result because it produces a discriminating question. Translation failure can protect a tradition from being conscripted into a problem it was not designed to solve. Metaphysical remainder can mark the exact boundary between an experiment and an ontology.

Keep four evidence ledgers apart

A comparison can fail even when its concepts are careful if its evidence is mixed without notice. A tradition-internal ledger records what a text or lineage claims and what practice the claim serves. A historical ledger records authorship, translation, development and disagreement. A structural ledger states the relation proposed across maps. An empirical ledger records what a controlled observation supports. None automatically settles the others.

Suppose trained practitioners report that the bell is present without an owner. The report is evidence for a reproducible description if elicitation and sampling are sound. It may strengthen a phenomenological model of reduced ownership. It does not prove that early Buddhist texts meant exactly that report, nor that an enduring self is metaphysically absent. Conversely, a faithful textual reconstruction does not show that contemporary practitioners instantiate the described sequence. The ledgers can constrain one another, but only through an explicit bridge.

This separation also protects consciousness-primary research. If awareness is fundamental, a tradition’s ontology may be a serious hypothesis rather than decorative language. Yet the empirical ledger still needs a discriminating observation. Treating awareness as potentially primary raises the explanatory burden on physicalist and functional accounts; it does not convert contemplative testimony into a mechanism measurement. The matrix can host the metaphysical claim without pretending that its current intervention has decided it.

Internal diversity creates a second boundary. “Buddhist”, “Indian”, “Chinese” and “phenomenological” are library shelves, not unitary theories. Buddhist traditions disagree about momentariness, reflexive awareness and the status of consciousness. Indian debates disagree about self, awareness and valid cognition. Chinese debates disagree about human nature, naming and the reliability of distinctions. Phenomenologists disagree about the ego, noema, embodiment and naturalisation. The matrix must attach every record to a text, lineage, interpreter and scope. Otherwise the very tool designed to prevent flattening becomes a new flattening device.

The seam is not a defect in convergence research. It is the location of its discriminating power. Where two maps oppose one another, design a test if the opposition has observable consequences. Where they do not, preserve the remainder instead of manufacturing a verdict.

Build comparisons that can fail

A rigorous comparison begins with reconstruction, not synthesis. Write each map without borrowed vocabulary. State its practical aim, its unit of analysis, its permitted evidence and its internal controversy. Only then choose a shared episode and a variable that can be changed. The comparison starts where both records make a claim about that change.

The workflow has four gates. The reconstruction gate rejects a map that its own specialists would not recognise. The intervention gate rejects a bridge with no shared counterfactual. The inference gate types what the result can support, from local functional agreement to unresolved ontology. The revision gate records nulls and break conditions instead of rescuing the preferred convergence story.

Apply the same discipline to artificial systems

The method matters beyond historical comparison. An artificial system may produce reports about attention, selfhood or witnessing. The temptation is to align those reports with a traditional map and infer a common architecture. The relation record asks harder questions. What system boundary produced the report? Which state was available? What intervention changes it? Which typed action follows? Does an independent readback verify the claimed change?

For example, a model may say that it “noticed an impulse and let it pass”. That sentence could be generated from prompt context without any persistent state, control veto or changed action policy. A configured agent might instead maintain an explicit proposal, apply an authority rule, abstain from execution and store a decision receipt. The two outputs sound alike, but only the second has an inspectable restraint path. Even then, the path establishes governed control, not felt non-attachment.

The purpose rule blocks another common error. Model checkpoints, episodic traces, enterprise knowledge, world state and evidence ledgers should not all be mapped to a tradition’s broad memory concept. They carry different authority and fail under different interventions. A useful bridge names one relation, such as disposition shaping later selection, and tests it while recording the break. Machine comparison becomes credible when operational state and authority are typed before philosophical language is added.

The lab encodes a deliberately conservative rule. A positive result permits a local bridge around the selected intervention. It does not permit a claim that two traditions share a complete architecture. A conflict result permits a discriminating comparison if the opposition has observable consequences. Translation failure means the records need reconstruction, not that the traditions are irrational or incomparable. Metaphysical remainder means the operational bridge can proceed while ontology stays open.

The tool cannot establish consciousness, spiritual realisation, historical influence or the truth of a tradition. It does not replace language expertise, practice knowledge or source criticism. Its value is narrower: it makes the analyst declare what is preserved, what breaks and what evidence would change the verdict.

from dataclasses import dataclass, asdict
from typing import Literal

Relation = Literal["align", "partial", "oppose", "unknown"]

@dataclass(frozen=True)
class Comparison:
    trigger: Relation
    transformation: Relation
    consequence: Relation
    purpose: Relation
    exclusion: Relation
    ontology: Relation


def classify(record: Comparison) -> tuple[str, str]:
    values = asdict(record)
    core = [values[name] for name in
            ("trigger", "transformation", "consequence")]

    if sum(value == "unknown" for value in core) >= 2:
        return "translation failure", "No shared intervention-sensitive core is specified."

    if "oppose" in core or values["purpose"] == "oppose" or values["exclusion"] == "oppose":
        return "conflict", "A common question exposes opposed causal or practical claims."

    ontology_open = values["ontology"] in {"oppose", "unknown"}
    fully_aligned = all(value == "align" for value in core)
    purpose_aligned = values["purpose"] == "align"
    exclusion_aligned = values["exclusion"] == "align"

    if fully_aligned and purpose_aligned and exclusion_aligned:
        if ontology_open:
            return "metaphysical remainder", "The operational bridge holds while ontology remains undecided."
        return "agreement", "The selected relation survives at a compatible grain and purpose."

    suffix = " Metaphysical remainder also remains." if ontology_open else ""
    return "qualified correspondence", "A local relation is shared, but grain, purpose or exclusion differs." + suffix


if __name__ == "__main__":
    example = Comparison(
        trigger="partial",
        transformation="partial",
        consequence="align",
        purpose="oppose",
        exclusion="oppose",
        ontology="oppose",
    )
    verdict, reason = classify(example)
    print(f"{verdict}: {reason}")

stateDiagram-v2
  [*] --> Claim
  state "Why the architecture metaphor<br/>arrives too early" as Claim
  state "A comparison record with break<br/>conditions" as Test
  state "Five outcomes, not one<br/>similarity score" as Revision
  Claim --> Test: expose mechanism
  Test --> Decision: evidence survives
  Test --> Revision: rival survives
  Revision --> Claim: narrow or rebuild
  Decision --> [*]

Build only the bridge whose relation survives intervention

When a researcher, engineer or philosopher encounters two traditions using apparently similar language, the old decision is to choose between universalism and incommensurability. Universalism says the maps must converge on one hidden architecture. Incommensurability says their worlds are too different for comparison. Both decisions arrive too early.

The better decision is to ask for a shared intervention. Reconstruct each map in its own terms. State the practical aim that selected its units. Replace translated nouns with typed relations. Change one causal property of a common episode and record what each map predicts will remain, disappear or reverse. Then classify the result without forcing it into a single scale.

This changes how a comparative research programme allocates effort. Do not spend the next study searching for a neural correlate of “mind” across traditions. Test whether learned naming changes temporal anticipation while preserving sensory discrimination. Do not treat a report of witnessing as proof of an independent witness. Separate the report’s functional and phenomenological structure from the ontology it is taken to reveal. Do not merge cultivation, liberation and description into one optimisation target. Preserve purpose in the experimental design.

The practical rule is simple: build a bridge only across relations that survive a named intervention, and write the break condition onto the bridge. An agreement then becomes usable. A conflict becomes testable. A translation failure sends the analyst back to the sources. A metaphysical remainder stays open without being dismissed.

The four maps around the bell need not become one atlas. Their partial overlaps can still guide research, precisely because the blank spaces and incompatible borders remain visible. The decision is no longer whether the traditions are secretly identical. It is which local relation has earned the right to travel.