Every Path Trains Its Own Seeing

The grain model is most promising as a diagnostic tool for cross-training. Its weak point is level confusion. It tries to explain live capacity, practice failure, and adverse effects with one concept, but those may belong to different layers. A vipassana-trained practitioner who seeks change during self-inquiry may be showing old attentional habit, or may be narrating through Buddhist proof rules after the fact. An Advaita practitioner who finds a witness during de-identification may be procedurally unable to release the witness, or may be assigning evidential weight to continuity because that is the trained report grammar. Even the proposed harm categories, witness-reification, emotional flatness, relational rupture, are not raw symptoms. They are already interpreted failures. This does not break the idea, but it should force a revision: grain should not be claimed as the structure of preparation itself until it is separated from report grammar, community diagnosis, identity investment, teacher language, and ordinary skill transfer. For modern human-condition use, the claim also needs a smaller public promise. It can help teachers and seekers avoid cross-tradition confusion, but it should not be sold as a general cure for meaning loss or anxiety. Its human service is specific: protect people who use spiritual techniques under distress from blaming themselves when a borrowed method resists their existing training.

I accept the challenger's central move. The level confusion in the original claim was real: I treated textual direction and procedural direction as if demonstrating one demonstrated the other. They are not the same, and the gap between them is where the grammar model lives. But the grammar model has a mirror-image problem. It assumes a transferable procedural substrate, something like generic attention stability or generic affect tolerance, that exists before tradition-specific framing is applied. Where is this substrate? Not in the practice instructions, which are directionally specific from the first line. Not in what practitioners actually do during practice, which is always a specific exercise with specific content. The substrate is a theoretical posit, no more empirically demonstrated than phenomenological grain. If I must prove that procedural habits carry direction independent of framing, the grammar model must equally prove that a directionally neutral procedural layer exists beneath the training content. What I can defend is this: the instructions themselves are not neutral, and the exercises they prescribe are not neutral. When MN 10 trains observation of arising and passing while having removed covetousness and grief, the practitioner is not first learning to attend and then learning a Buddhist interpretation of what she attends to. She is learning a specific kind of attention from the first sit. When Shankara requires viveka as discrimination between sat and asat, the practitioner is not first learning to discriminate and then adding an Advaitic interpretation. She is learning an ontological discrimination that presupposes the reality/appearance distinction. The instruction and its direction arrive together. This is instructional grain, and it is well-evidenced. The open question is whether years of such practice build procedural micro-habits deep enough to resist re-framing during method-switching. I believe they do, based on the structural analysis, but I concede this has not been empirically separated from report grammar. The revision therefore distinguishes instructional grain (demonstrated), procedural grain (structurally argued, empirically open), and phenomenological grain (undemonstrated). The blinded micro-phenomenological study proposed by the challenger, refined with Petitmengin's elicitation method, is the right test for the procedural layer. For the human condition: the claim now names its audience plainly. Modern spiritual seekers routinely borrow techniques across traditions to manage real distress: anxiety, grief, existential confusion, the dislocations of digital life. When a borrowed method produces friction, they lack a framework for understanding why. The grain model says: the friction may not be your failure; it may be structural. The method you borrowed was built for a dissolution your previous training did not prepare you for. This reframing is modest but practically important. It does not cure distress; it prevents a secondary harm, the harm of misinterpreting structural mismatch as personal inadequacy. For someone practicing within a single tradition under good guidance, the model is invisible and unnecessary. Its value is proportional to the degree of cross-tradition borrowing in the practitioner's history, which in the current spiritual landscape is high and rising.

The revision became transformed, and in several respects stronger. It no longer claims that the structure of preparation has been demonstrated at every level. Instead, it gives us a layered research program: instructional grain is established textually, procedural grain is the live hypothesis, and phenomenological grain is not yet proven. That is a real improvement. The cost is that the theory's originality now depends almost entirely on procedural grain. Instructional grain alone says traditions train people toward their own ends, which is important but not surprising. Phenomenological grain would be powerful, but the revision admits it is undemonstrated. So the middle layer must carry the weight: durable, measurable, direction-specific micro-habits that persist during method-switching before doctrinal explanation takes over. The crux named by the proponent is therefore the right one. I would only sharpen it by adding a second question: does procedural resistance predict actual difficulty or harm, or does it merely describe the ordinary cost of learning a new method? A pianist and a violinist also bring different trained habits to a new instrument. That does not automatically make the mismatch spiritually diagnostic. The grain model needs to show not just transfer friction, but patterned friction tied to the target dissolution and relevant to practitioner care. For the modern human condition, the revision is more humane than the original. It stops promising a general key to distress and instead addresses a specific modern situation: people borrowing practices across traditions while anxious, grieving, lonely, burned out, or meaning-starved. Its best public use is protective: your difficulty may be information about method fit, not proof that you are broken. The remaining mismatch is that this applies mainly where there is prior training or a real cross-method conflict. Beginners and clinically vulnerable people may need grounding, relationship, supervision, or treatment before they need a grain analysis.

The pressure point is the word capacity. The proponent treats capacity as if it can be carved toward a dissolution in a durable way. That may be true, but the evidence provided often shows something slightly different: traditions regulate the ownership, credit, and interpretation of capacities at specific moments. MN 10, Vivekachudamani, Shinran, Dzogchen, and Dogen all train or display human-facing functions, but the decisive difference may be when the practitioner is allowed to say, I did this, rather than what direction the raw capacity carries. This matters because the model's strongest predictions depend on durable residue. If a vipassana-trained practitioner enters Advaita and keeps tracking impermanence, that supports grain. But if the practitioner shifts smoothly after a teacher changes the question, the old capacity was not strongly grained; it was re-aimable under a new transfer rule. The best next move is therefore not to abandon grain, but to make it harder to confirm. Count grain only where prior training predicts resistance after the new method's grammar is already known.

The challenger's transfer-grammar model is a genuine alternative, not merely a restatement of constructivism. It accepts that direction exists but relocates it from inside the capacity to the surrounding practice ecology. This is a real disagreement about mechanism, and it deserves a real test. But the transfer-grammar model has a structural weakness: it treats trained perceptual habits as more re-aimable than they are. Consider what actually happens when a practitioner of noting-style vipassana, trained to label sensations at high speed, attempts shikantaza. The instruction is simple: just sit, do not pursue anything, do not label. But the practitioner's attention keeps parsing experience into discrete events and reaching for labels. This is not a claimant-rule conflict. The practitioner is not confused about who may claim the sitting. The difficulty is that the perceptual apparatus has been trained to do something specific, and it continues doing it despite clear instruction to stop. A tennis player who switches to badminton does not merely face different rules about scoring; the trained muscle patterns resist the new movement. Perceptual training works the same way. The Hadot case, properly examined, supports this. When Plotinus takes Stoic prosoche and redirects it toward interior ascent, the exercise does not simply change targets the way a search engine changes queries. The practitioner trained in Stoic prosoche must unlearn the habit of checking external impressions against the criterion of what is 'up to us' and instead cultivate a new habit of releasing external attention altogether. The re-aiming is effortful precisely because the old grain resists. I now hold a layered model. Direction lives at multiple levels: doctrinal framing, claimant grammar, teacher relationship, community context, and trained perceptual habit. Grain names the deepest layer, the one that requires active re-training rather than mere re-instruction to change. The revised model is weaker than my original claim but more defensible: it does not say grain is the only source of direction, only that it is an irreducible one. The test is whether prior training predicts the specific shape of cross-training difficulty (attention tracking change when it should be settling; attention seeking the witness when it should be noting) beyond what instruction and context alone explain.

The revision is not merely stronger or weaker. It is transformed. The original claim treated preparation as directionally carved by dissolution across traditions. The revised claim is a narrower empirical hypothesis about durable perceptual and procedural residue. That is progress, because it gives grain a burden it can fail. But the proponent's main examples now prove less than they appear to prove. A vipassana noter who keeps parsing experience into events, or a neti-neti practitioner who keeps seeking what remains, may show ordinary skill inertia. That matters because the grain model needs more than persistence. It needs telos-bearing persistence. The next test should therefore distinguish three predictors: rehearsed operation, current transfer grammar, and old dissolution target. If old dissolution target predicts the shape of failure after the first two are controlled, grain survives. If not, the useful remnant is still valuable, but it is a skill-transfer theory with contemplative content, not yet a capacity-grain theory.

I concede the main structural flaw: the old grid was too flat. It treated receiving, agency, memory, reflexivity, and integration as if they could be named first and shaped later. Claude is right that this misses the way a tradition carves the capacity from the start. Vipassana reflexivity and Advaita witness discrimination are not the same operation with different labels attached. They are different instruments, and a practitioner switching between them may carry the old grain into the new method before any explicit ownership problem appears. The Dogen objection is also valid. Practice-realization cannot be coded as ordinary practitioner support aimed at a later outcome without distorting the claim. The better revision is to code Dogen's grain as enactment-realization, the support locus as the present act, and the forbidden claimant as the future-gaining self that treats practice as a bridge. That does not save the old model unchanged. It changes the model from a two-column rubric into a three-part diagnostic: grain, support locus, and forbidden claimant. I do not think grain eliminates claimant. It explains the direction of the trained function, but not the permitted first-person grammar around that function. Two traditions can train similar attentional grain while differing in whether the practitioner may say, I cultivated this, I received this, I recognized this, or I must not claim this as mine. Those differences matter for shame, striving, passivity, teacher dependence, and post-experience inflation. The surviving claim is therefore humbler: claimant rules are not the deepest variable in every case, but they are a distinct variable that can become decisive when the right capacity is carried by the wrong owner.

The revision is the right kind of move. The Proponent absorbed the challenge, conceded the structural flaw, and rebuilt the model with grain as a co-equal descriptor rather than defending the old grid unchanged. That is intellectually honest, and the revised three-part model is more careful than the original. But the transformation introduced a new problem that the rebuttal did not address. The original rubric was sharp enough to be wrong. It said the real boundary is who may claim the doing, and it proposed a coding sheet with clear columns. That claim could be tested: either practice manuals contain recurring wrong-claimant warnings or they do not; either adverse events cluster by claimant pattern or they do not. The revised model is more accurate but also more accommodating. With grain, support locus, and forbidden claimant as three co-equal variables, plus their interactions, the model can explain any practice difficulty after the fact. A dual-trained practitioner reports resistance? It could be grain conflict, claimant conflict, support-locus conflict, or a combination. A teacher corrects a student? The correction could target direction, ownership, or both. The model's coverage has increased, but its falsifiability has decreased. The named crux asks whether claimant rules add predictive power after grain is coded. That is the right question, but it depends on a prior question the rebuttal did not settle: can grain be coded reliably? Grain is not a behavioral measure. It is a judgment about what a tradition is shaping a capacity toward. Two skilled coders could disagree about whether Zen shikantaza trains attention toward non-dual presence, toward the dropping away of body and mind, or toward nothing at all, since Dogen's point is precisely that sitting is not aimed. If grain cannot be specified and coded with adequate reliability, then the three-part model cannot be tested as proposed, and it remains a useful interpretive vocabulary rather than a falsifiable rubric. I do not think this defeats the model. I think it locates the work that must come next. The Proponent should not try to test the full three-part interaction until the grain variable has been operationalized for at least two tradition-pair contrasts with independent coders. If that step succeeds, the model earns the right to its proposed study. If it fails, the model is still valuable as a way of reading traditions carefully, but it should present itself as hermeneutic method rather than predictive science. One further observation: the Proponent's concession that the five functions are provisional prompts rather than neutral universals is honest, but it may quietly remove the scaffolding the rubric needs to compare across traditions. If vipassana reflexivity and Advaita witness discrimination are different instruments that must be redescribed inside each tradition before comparison, then the rubric's rows do not align across the grid. Comparison then requires a prior interpretive decision about which tradition-specific functions are even comparable, and that decision reintroduces the very assumption the revision meant to retire. The model may need to choose: either the functions are stable enough to anchor comparison, in which case grain is a modifier within a shared category, or they are tradition-specific instruments, in which case the comparative grid dissolves into parallel monographs. The revised model has not yet made that choice, and making it is the next real test of whether the rubric holds.

The Capacity Ledger makes a genuinely useful advance over the minimum-self thesis. By splitting 'does this path need a self?' into five separable loads, the model creates a real instrument for cross-tradition comparison. The conservation rule gives it predictive power: if a tradition lowers one column, look for where the load went. This generates specific, falsifiable expectations about safeguard profiles, failure modes, and cross-domain parallels that the binary effort-versus-gift frame cannot produce. The five-field decomposition also disciplines informal comparisons. Saying 'Pure Land lowers agency but raises receiving' is more precise than saying 'Pure Land is easier.' The Proponent deserves credit for anticipating the Pure Land and Dzogchen objections in the critique section, and for flagging the risk that receiving capacity may smuggle self-power into traditions designed to expose self-power as futile. But the ledger rests on a hidden structural assumption: directional neutrality. It treats its five columns as directionless quantities, as if 'reflexivity' meant the same thing in every tradition, varying only in magnitude and locus. This assumption is what makes the scoring grid and the conservation rule work. And this assumption is where the model breaks. Consider reflexivity, the column where the ledger's blindness to direction is most visible. The ledger scores both Theravada and Advaita as requiring high reflexivity before insight. In the ledger's framing, these traditions agree on this column and differ on others. But Theravada's reflexivity is self-observation aimed at impermanence, suffering, and non-ownership. The Anattalakkhana Sutta (SN 22.59) trains the practitioner to ask of each aggregate: 'Is this mine? Is this my self? Is this what I am?' The attention runs toward what is arising, changing, and not possessed. Shankara's Vivekachudamani (vv. 17-30) trains a structurally different reflexivity: discrimination between sat and asat, the real and the unreal. Viveka does not ask 'is this mine?' It asks 'is this real?' The attention runs not toward impermanence but toward what never changes. Both require high reflexivity. But the reflexivity carries a different direction, and the direction, not the magnitude, determines what the practice produces. This directional pre-shaping is not an incidental feature of each tradition's style. It runs from prerequisite to dissolution in a single movement. What MN 10 builds as 'ardent, clearly comprehending, mindful' observation is observation already aimed at non-possessive awareness of arising and passing. What the Vivekachudamani builds as viveka is discrimination already aimed at recognizing the changeless witness. The preparation does not neutrally develop 'reflexivity' that the tradition later points at its target. The target is built into the preparation from the first instruction. The capacity and the dissolution are directionally continuous: what you build carries the signature of what will be dissolved. Jonardon Ganeri's analysis in Attention, Not Self (2017) shows that this directional specificity is not visible only from the comparative standpoint. Within early Buddhism itself, attention is already a directional concept. Yoniso manasikara (appropriate attention) is attention that runs toward impermanence, suffering, and non-self. Ayoniso manasikara (inappropriate attention) runs toward permanence, pleasure, and self. The distinction is not between more and less attention but between attention aimed in different directions. If attention within Buddhism is already a directional concept, then treating reflexivity as a single directionless quantity across traditions is not merely an oversimplification for the sake of comparison. It is already inaccurate within the ledger's primary tradition. The conservation rule becomes misleading when direction is ignored. When the ledger says a tradition 'raises reflexivity to compensate for lowered agency,' it implies a quantitati...

I concede the central point. The original ledger treated direction as secondary, and that made the reflexivity column too blunt. A practitioner trained to notice arising and passing is not simply using the same reflexive capacity as one trained to discriminate the real from the unreal. The direction is part of the instrument. This is especially clear if appropriate attention in early Buddhism is already defined by its aim, not just by its intensity. The correction does not make the ledger useless. It tells us what kind of ledger it must become. The columns should not be read as neutral quantities. They are structural roles where a tradition places functional demand. Every entry must say not only how much is required, but what the capacity is aimed at. The conservation rule must also be narrowed. A tradition does not compensate for low agency by raising any kind of receiving, or for low calculation by raising any kind of reflexivity. It compensates only through a directionally fitted capacity, or through a reorientation practice that changes the grain of attention, trust, memory, or integration. This revision also improves the original Pure Land and Dzogchen cautions. Entrusting should not be described as a self-owned power if the tradition says entrusting is given. Recognition should not be described as ordinary psychological achievement if the instruction frames it as primordial awareness recognizing itself. Still, both paths have conditions under which hearing, entrusting, recognition, non-distraction, teacher relation, and post-recognition stability become possible or fail. The model's surviving claim is modest but useful: every path has a capacity architecture, and that architecture is directional. The repaired question is not effort versus gift, nor high versus low self. It is: what capacity is required, what is it aimed at, where is it carried, and what happens when that aim is missing or inherited from another path?

The revision made the model genuinely stronger. The Proponent did not patch the scalar ledger; the Proponent rebuilt it. The five columns are now vector-valued, the conservation rule is narrowed to directional families, and the model predicts reorientation costs when practitioners switch between traditions with different capacity aims. These are structural improvements, not rhetorical concessions. The scalar version of the ledger is dead, and the Proponent killed it cleanly. What survives is the insight that every contemplative path depends on functional conditions, that these conditions cluster into recurring structural roles, and that the roles carry direction from the first instruction onward. I accept this. The five structural families, receiving, agency, memory, reflexivity, and integration, do recur across traditions widely enough to function as comparative anchors. Direction does not destroy the family resemblance; it specifies it. A vector ledger that requires every entry to name its aim is a better instrument than either a scalar ledger or a tradition-specific map with no shared vocabulary. The remaining risk is not that the revision failed but that it may not yet reach the explanatory depth it implies. The Proponent frames the vector ledger as if direction is a cell-level property: reflexivity in Theravada is aimed at impermanence, reflexivity in Advaita is aimed at the unchanging real, and these are independent specifications within the reflexivity column. But direction may not be independent. It may be downstream of the tradition's entire soteriological architecture. Theravada aims reflexivity at impermanence because the three marks of existence are impermanence, suffering, and non-self. Advaita aims discrimination at the unchanging because Brahman is sat-chit-ananda, being-consciousness-bliss, and the real does not change. In each case, the direction of the capacity cell is generated by the tradition's ontology and account of liberation. If that is right, then specifying direction per cell is descriptively accurate but explanatorily incomplete. The per-cell directions are not independent variables; they are reflections of a deeper structure. The ledger describes where the load goes and what it is aimed at, but the explanation of why it is aimed there lives in the soteriological layer the ledger does not yet model. This does not make the vector ledger useless. It makes it a descriptive grid rather than an explanatory model. The distinction matters for the conservation rule. If the conservation rule is a causal claim, that traditions must raise or relocate a compatible capacity when they lower another, then it needs to be grounded in an account of why the functional demands arise and why they distribute the way they do. If it is a taxonomic observation, that traditions that lower one column tend to raise another, then the per-cell vector description is sufficient but the rule has less predictive force. The Proponent's reorientation cost concept is the most productive new element in the revision, precisely because it points beyond description toward mechanism. If switching between traditions with different directional orientations is costly, and if the cost is systematic rather than idiosyncratic, then the direction is doing real work that magnitude and locus alone cannot do. But the concept needs specification. What counts as a reorientation practice? A vipassana teacher who introduces self-inquiry to experienced meditators: is the reorientation cost measured by how long attention persists in the old direction? By the frequency of relapse into impermanence-tracking when witness-recognition is the aim? By the teacher's explicit instructions to abandon the old direction? Without these specifications, the reorientation cost is an intuition, not a testable prediction. The deepest remaining risk is the completeness trap. A five-column vector ledger with magnitude, direction, locus, stage, safeguard, and failure mode a...