一、這個概念在說什麼
In the Buddhist logical tradition, particularly as articulated in 《正理滴論》 (*Nyāyabindu*, as translated from the Tibetan — a core text of the Dharmakīrtian 因明 / *Hetuvidyā* lineage), a valid inference (量式 / *pramāṇa*) is structured around three members: the proposition to be established (宗 / *pratijñā*), the reason (因 / *hetu*), and the example (喻 / *dr̥ṣṭānta*). The reason must satisfy three conditions known as 因之三相 (*tr̥ṇirūpa*) — it must be present in the subject of the proposition, present in similar instances, and absent in dissimilar instances.
隨因後行 (literally "following the reason, going forward") names the determinate forward concomitance between a reason and the proposition it is meant to establish. Wherever the reason reliably appears, the proposition reliably follows. It is paired with 隨因遣行 ("following the reason, going away / excluding") — the determinate backward exclusion: where the reason is absent, the proposition is also absent. Together they are the two faces of a single logical requirement, examined from the positive side and the negative side.
The classical definition carries weight that the ordinary translation can hide:
凡能詮表某種相系屬者，即名為「隨因後行」。
"Whatever can express a kind of *determinate correlation* (相系屬) is named 隨因後行." The word *determinate* is the load-bearing piece. 隨因後行 is not mere co-occurrence; it is the kind of correlation that has a structural basis in either 自性關係 (intrinsic-relation / *svabhāva-sambandha*) — where one thing is itself the nature of the other, as a definition is to what it defines — or 因果關係 (cause-effect / *kārya-kāraṇa-sambandha*) — where one thing is the cause or effect of the other. If two things are not related in either of these ways, the text warns, removing one will not necessarily remove the other, and there is no 隨因後行 to speak of.
The text then spells out how the two concomitances interact with the parts of the syllogism:
- Through the 同品 (similar instance), the practitioner discerns that the reason follows in the direction of the proposition (隨因後行): if the reason is absent in the proposition, there is no 隨因後行. - Through the 異品 (dissimilar instance), the practitioner discerns the reverse, the exclusion (隨因遣行): if the proposition is absent, the reason is absent; if the reason is absent, the proposition cannot be established.
A subtle but important point: the text notes that one formulation alone suffices. The single phrase "隨因後行", by its inferential force, can display the presence of the reason in similar instances and its absence in dissimilar ones. One does not need to lay out both faces rigidly in every argument.
The relationship between the two is rigorous. Their natures are mutually exclusive (互相排除): if one is determinately absent, the other is determinately present. If both remain uncertain, the inference is 不定因 (indeterminate reason) — it cannot decide the matter, and the proposition is not established.
The 因之三相 三相 (three marks of valid reason) framework is the technical home of these two concomitances. 隨因後行 primarily concerns 同品定有性 (definite presence in similar instances); 隨因遣行 concerns 異品遍無性 (universal absence in dissimilar instances). When the reason is *absent* in similar instances and *present* in dissimilar instances, the reason is 相違因 (contradictory reason) — such as using "produced" or "effortfully originated" to prove "permanent."
The text also extends 隨因後行 to the 未緣到因 (non-apprehension) form of inference — a sophisticated move with contemporary application. Here, the underlying correlation is grounded in the structure of cognition itself: whatever exists must be capable of being apprehended (凡是有者，必為緣慮所明瞭). Hence, if a careful search has not apprehended a jar in a certain place, the jar is absent there. The positive-form (同品) inference affirms "apprehension"; the negative-form (異品) inference, by "non-apprehension," shows that what would be named by the relevant term is absent. This is not a license for careless negative inferences — it works only when the search is exhaustive and the only explanation for non-apprehension is genuine non-existence.
The fault vocabulary in the text names the four ways 隨因後行 can fail:
1. 無隨因後行 — the reason simply does not follow the proposition. Example: "All speakers are people with greed, etc." (凡言說者，皆為有貪欲等之人). Being a speaker does not determinately correlate with being greedy. 2. 未全示隨因後行 — the example only partially displays the correlation. 3. 隨因倒行 — the correlation is reversed. Example: "All impermanent things are produced" (凡是無常者，皆所作性). The real correlation runs the other way: produced things are impermanent, not impermanent things are produced. The reason is asked to carry weight in the wrong direction. 4. 相違因 — the reason is absent in similar instances but present in dissimilar ones. Forms of this were mentioned above.
The text's most pointed application is the dismantling of the inference of a self (我). When the Sāṃkhya or similar schools argue: "There is a self, because the life-principle is present in the body," Buddhist philosophers respond by showing that the proposed reason (life-principle) lacks 隨因後行 with the proposition (the existence of a self): in both possessing a self and not possessing a self, life-principle is present; for the proposition "a single self" it is also not established, precisely because there is no 隨因後行. The technical apparatus here is not academic ornamentation; it is the contemplative work of loosening the inferential structures that constitute the sense of self.
In the broader architecture of the Buddhist path (解脫道), 隨因後行 is one of the precision tools of pramāṇa / epistemology. Accurate reasoning is required for accurate understanding of phenomena — including the mistaken inference of self that keeps beings bound. Where the inference fails, the conclusion is not established; where the inference has determinate correlation, the conclusion can be carried. The framework is a hinge between the formal analysis of reasoning and the transformative work of liberation.
二、生活裡的走查
This is technical vocabulary, but the underlying experience is utterly ordinary. We run inferences constantly — about people, about ourselves, about the world. 隨因後行 names a question we can ask of every one of them: *does the reason actually carry the conclusion, or does it merely appear to?*
Scenario 1: The social media generalization.
A post circulates: "All CEOs are mentally unhealthy." The supporting reason implied: high-pressure roles produce mental illness.
- Proposition: CEOs are mentally unhealthy. - Reason: high-pressure work / status. - Does 隨因後行 hold? Is there a *determinate* correlation between being a CEO and being mentally unhealthy? No. Many CEOs are psychologically robust; many high-pressure workers in other roles are mentally unhealthy; many low-pressure workers are mentally unhealthy. The correlation is real but not *determinate* — it does not hold "wherever the reason is, the proposition follows." - Does 隨因遣行 hold? Are those who are not CEOs all mentally healthy? Obviously not. - Verdict: 不定因 (indeterminate reason). The inference does not establish the conclusion. This is the classical *hasty generalization* — and it is structurally identical to the text's example of "凡言說者，皆為有貪欲等之人." The correlation is between a subset and the broader category, not between the reasoned category and the proposed conclusion.
Scenario 2: The wellness claim.
A podcaster argues: "I took cold showers every morning for a year and never got sick. Therefore, cold showers prevent illness."
- Proposition: Cold showers prevent illness. - Reason: The podcaster's individual case. - 隨因後行? A single autobiographical case is not a determinate correlation. It is not a cause-effect relationship (因果關係) — cold showers and immunity are not intrinsically related (自性關係) in any obvious way. The "correlation" is one person's anecdote across one year. - 隨因遣行? Many people who took cold showers got sick; many who didn't take cold showers didn't get sick. - Verdict: an example of *post hoc, ergo propter hoc* — co-occurrence mistaken for causation. This is precisely the error the Buddhist framework guards against by demanding *determinate* correlation, not surface co-occurrence.
Scenario 3: The inference that actually works.
You arrive at a hilltop village. You see smoke rising from a distant valley. You form the inference: *there is fire down there.*
- Proposition: There is fire in the valley. - Reason: Smoke. - 隨因後行: Wherever there is smoke (in ordinary natural contexts), there is fire, because smoke is the effect of fire (因果關係). - 隨因遣行: Where there is no fire, there is no smoke (in this context). - Verdict: determinate correlation. The inference carries. This is the classical Indian example — and it works because the underlying relation is genuinely causal, not merely habitual.
Scenario 4: The 隨因倒行.
A manager argues: "All top performers in our company wake up before 6 AM. Therefore, waking up before 6 AM makes you a top performer."
- The proposition: Waking up early *causes* high performance. - The reason: Waking up early. - 隨因後行? Top performers do tend to wake up early — that part of the correlation is real. But the *direction* the manager is asking the reason to carry is the causal direction. The wake-up time is more plausibly a *symptom* of discipline, not the *cause* of performance. The reason is being asked to carry a causal conclusion, but the correlation as observed is correlational, not causal in the direction claimed. - This is the text's example of 隨因倒行 in modern clothing. The correlation exists; the direction is wrong. The text's example, "凡是無常者，皆所作性" (all impermanent things are produced), is structurally identical: the real correlation is "produced things are impermanent," but the statement asks the relation to run the other way and use impermanence as the predicate. The reason cannot carry the conclusion in the direction asked.
Scenario 5: The non-apprehension case.
You organize a quiet hour in a meditation hall. You do a careful sweep — looking, listening, attending — and find no sound. You conclude: the hall is silent.
- Proposition: The hall is silent. - Reason: No sound has been apprehended. - 隨因後行 / 隨因遣行: This works *only* because the search was exhaustive and the only explanation for non-apprehension (in this context) is genuine absence. If the search was sloppy, the inference fails. If sound could be present but unapprehended (e.g., infrasonic frequencies, sounds masked by tinnitus), the inference fails. - Verdict: a valid 未緣到因 inference, *under its conditions*. The framework does not license careless negative inferences ("I didn't see any aliens, therefore there are no aliens"), because the conditions of exhaustive search and reliable cognition are not met.
The most common slip in mapping this onto daily life: treating any perceptible correlation as a 隨因後行. The correlation is real — many CEOs are stressed; many cold-shower-takers are healthy for a year; many top performers wake up early. But the *determinate* correlation is not there. The Buddhist framework, in insisting on determinate correlation, is refusing to be fooled by the human mind's most natural error: mistaking the regular for the necessary.
The second slip: assuming 隨因後行 alone is enough. The mutual exclusion principle is strict: if 隨因後行 is unclear, 隨因遣行 is also unclear, and the inference is indeterminate. Both must be in place.
The third slip: forgetting the third mark (遍是宗法性 — the reason must actually be present in the subject of the proposition). A surprising amount of bad reasoning fails right here, before we even get to the concomitances.
三、當代人為什麼需要它
We live in an age of abundant reasons and scarce valid inferences. Every claim, every advertisement, every political argument, every internal anxious thought, arrives with reasons attached. The structural problem is not the absence of reasons but the failure of reasons to actually carry their conclusions. This is exactly the failure 隨因後行 diagnoses.
The framework is more fine-grained than formal logic (which checks the form of an argument, not whether the premises actually warrant the conclusion) and more demanding than casual "common sense" (which often rests on unexamined regularities). It asks: *what is the actual relationship between the reason and the proposition?* Is it an intrinsic identity — a definition, a nature-relation? Is it a real cause-effect relation, grounded in how things actually produce or arise from each other? Or is it a surface co-occurrence that doesn't actually carry the inferential weight claimed for it?
There are illuminating parallels with contemporary epistemology, though the parallels are structural, not claims of equivalence. Causal inference in modern statistics — Judea Pearl's framework, for instance — tries to formalize exactly what 因明 was grappling with: do we have a determinate causal relationship, or merely a correlation? Pearl's do-calculus asks whether an intervention on the cause produces the effect — which is operationally similar to asking whether the reason is determinately present in the proposition (隨因後行) and absent in its absence (隨因遣行). Modern Bayesian inference asks similar questions about the posterior probability of a claim given evidence — though it works in degrees rather than the binary determinate/non-determinate framework of Buddhist logic. The Buddhist framework does not have access to modern statistical machinery, but it is structurally demanding in similar ways, and arrived at similar requirements.
Cognitive science offers a parallel of a different kind. The human brain is, in Daniel Kahneman's terms, a "machine for jumping to conclusions" — it readily forms conditional expectations from observed co-occurrences. This is a survival engine, but it produces systematic errors: illusory correlation, the gambler's fallacy, the post hoc fallacy, the sunk-cost fallacy. Buddhist logic's insistence on *determinate* correlation (排除偶然共同出現) is precisely the corrective. Whether the brain's inferential architecture implements something like 隨因後行 behaviorally is an open empirical question; the parallel is suggestive, the proof is not yet there.
For Buddhist practitioners, the concept has a more direct and more transformative utility. The classical application is the dismantling of the inference of self (我) — the analytical work of pramāṇa in service of liberation. Every time we silently assert and operate on the inference "I am suffering because I am a self," "I am a self because I have continuity of memory," "I am a self because I have preferences" — we are running an inference. The 隨因後行 framework gives us the analytical tools to examine whether such inferences actually carry their conclusions, or merely *seem* to. Where they fail to carry, the sense of self loosens, not through violence but through precision. This is the contemplative use of the apparatus.
In an age saturated with information, the capacity to evaluate whether a reason *actually carries* its conclusion — and not merely appears to do so — is less a philosophical refinement than a survival skill. 隨因後行 is one of the most precise tools the Buddhist tradition offers for this work. It serves the household reasoner as a basic literacy; it serves the practitioner as a contemplative instrument.
四、常見誤讀與澄清
Misreading 1: 隨因後行 is just "whenever both are present, the inference holds."
This is the surface-co-occurrence confusion. The text is explicit: "隨因後行並非僅指表面的同時出現" — it is not merely surface co-occurrence. It requires a determinate correlation (可確定的相系屬), grounded in either intrinsic-relation or cause-effect. The CEO-and-stress example illustrates this: CEOs and stress often co-occur, but the correlation is not determinate. Accidental or partial co-occurrence is not 隨因後行.
Misreading 2: 隨因後行 is the same as 同品定有性.
Closely related, but not identical. 隨因後行 is the general logical concept of the forward concomitance; 同品定有性 is its particular technical expression within the three-marks structure. The text explains that 隨因後行 works through the determination of similar instances, but the broader concept includes the determination by either intrinsic-relation or cause-effect, and the 在同品法/異品法 of the syllogism — not strictly the second mark alone.
Misreading 3: 隨因後行 alone is sufficient.
Strictly false. 隨因後行 and 隨因遣行 are mutually exclusive but jointly required for a valid inference. The failure of either is the failure of the inference. The text's example of an 不定因 (indeterminate reason) is precisely this case: 隨因後行 is unclear, 隨因遣行 is unclear, the proposition is not established. Both must be in place.
Misreading 4: 隨因後行 is a Buddhist invention of purely historical interest.
The structural requirement (a reason must co-occur with its proposition and be absent in its absence) is a universal requirement on inference. Modern formal logic, Bayesian inference, and causal modeling all grapple with the same structure. The specifically Buddhist contribution is the insistence on determinate correlation (not just correlation), the two-fold categorization of correlations into intrinsic-relation and cause-effect, and the practical application to the dismantling of the self-inference. The apparatus is technical, but its requirements are not sectarian.
Misreading 5: 隨因後行 validates induction or frequentist correlation.
A subtle error. The phrase "wherever the reason is, the proposition follows" sounds like a generalization — but the Buddhist framework is not committed to enumerative induction. 隨因後行 is grounded in a *determinate* correlation by nature or by cause-effect — a stronger claim than mere Humean regularity. This is closer to what contemporary philosophy of science calls a *lawful* or *nomic* generalization, not a frequency-count. Hume would not recognize it as knowledge; Dharmakīrtian logic would.
Misreading 6: 隨因倒行 is a different kind of error from 隨因後行.
It is not a different kind of error; it is the inversion of 隨因後行. The correlation exists but is being read in the wrong direction. The text's example — "all impermanent things are produced" — is the statement of a real correlation, but with the roles reversed. When such a statement is used as a reason for some proposition, the reason runs backward and fails to carry the inference. To label 隨因倒行 as a generic "contradiction" misses the structural point: the reason is asking the correlation to run in the wrong direction.
Misreading 7: 隨因後行 is purely philosophical and has no practical application.
This is the most unfortunate misreading, because the classical application — the dismantling of the inference of self — is one of the most important practical exercises in the Buddhist contemplative life. Every time we silently assert "I am suffering because of X," "I am a self because I have memories," "I am a self because I have continuity," we are running an inference. The 隨因後行 framework gives us the tools to examine whether these inferences actually carry their conclusions or merely *seem* to. The framework is not only descriptive of reasoning; it is intervention in the inferential habits that constitute the sense of self. Taking it as merely academic is to miss the contemplative work it is designed to do.
結語
隨因後行 names a precise epistemic requirement: that the reason actually determinately correlates with the proposition-to-be-established, not merely appears to do so. It is half of the conjoint requirement whose other half is 隨因遣行. Together — and only together — they form the structural backbone of valid inference in the Buddhist logical tradition, particularly as articulated in《正理滴論》.
The term is technical, but its substance is not arcane. It is the very thing that distinguishes an argument that carries its conclusion from one that merely waves in its direction. In an age where every claim arrives with reasons attached, the capacity to discern which reasons actually carry their conclusions is not just a philosophical refinement. It is a basic form of intellectual hygiene. And at its deepest level, it is a tool for the dismantling of the very inferential structures that constitute our sense of self — structures we have been running, mostly unexamined, since long before we had any reason to think of them as "inferences" at all.