What This Concept Is Saying
Suppose a friend says, "Sound is impermanent because it is produced." You can nod along — or you can ask the question that Buddhist logicians spent centuries sharpening: *produced-ness also shows up in lots of things that aren't impermanent, or that aren't relevant here. Does it actually pin down impermanence?* The technical term for the set of cases that lets you test that question is **異品 (*vipakṣa*)** — "the different class," the collection of instances in which the property you are trying to prove (所立法, *sādhya-dharma*) is absent.
The 《因明入正理論》 (*Nyāyapraveśa*, T1630) defines it in one taut line that is doing a lot of work: 「異品者，謂於是處無其所立」 — "The dissimilar class is that in which the property to be established is not found." And the 《正理滴論》 (*Yuktiṣaṣṭikā*'s commentary tradition, B0042) puts the same idea differently: 異品 is *"非同品"* — "whatever is not the similar class" — that is, whatever does not match the property being proven.
To see why this matters, we need the whole three-legged frame of 因三相 (*trairūpya*, "three characteristics of a valid reason"):
1. **遍是宗法性 (*paksadharmatā*) — the reason must actually be a property of the subject (e.g., "produced" must be true of sound). 2. 同品定有性 (*anvaya*) — the reason must be present in at least some instances similar to the predicate (e.g., "produced" is true of bottles, pots, lightning — things we agree are impermanent). 3. 異品遍無性 (*vyatireka*) — the reason must be universally absent in all instances dissimilar to the predicate (e.g., "produced" is not** true of empty space, 虛空, which is held to be permanent).
異品 is the territory where the third leg is tested. *Not "where the opposite is present," not "where things are bad," not "where similar things fail"* — strictly: where the property to be established is absent. This precision is the whole point. If you can find even one instance in this class where the reason still shows up, the reason is no longer decisive.
Why the asymmetry between legs 2 and 3?
One thing often puzzles newcomers: leg 2 says "definitely present in some similar instances" (not *all*), but leg 3 says "universally absent in all dissimilar instances." Why the difference?
The classical answer is that leg 2 is inductive — you only need a representative sample of similar instances to show co-occurrence. Leg 3 is exclusive — even a single counterexample in the dissimilar class means your reason could also be tracking something *other than* the property you're trying to prove. So the bar on 異品遍無性 is higher: total absence, not partial absence.
The Three Subtypes from the 《正理滴論》
The 《因明入正理論》 treats 異品 as a single, undifferentiated class. The 《正理滴論》 is more analytic and splits 異品 into three:
1. 「包括彼外之異品」 — the dissimilar class that "includes beyond it": the broader surrounding domain that goes past the similar class. 2. 「與彼相違之異品」 — the dissimilar class that "contradicts it": instances that actively oppose the property to be proved. 3. 「彼上沒有之異品」 — the dissimilar class "on which [the property] is absent": instances where the property simply does not appear, without actively contradicting it.
In plain English: when you say "things where impermanence is absent," you might mean (a) the whole surrounding world beyond impermanent things, (b) things declared permanent like 虛空, or (c) things where impermanence just doesn't apply in any obvious way. In practice, all three end up tested by the same third leg of 因三相, but the split matters when you are refining a tight argument and need to be clear about which kind of counter-instance would actually defeat it. The 《因明入正理論》 doesn't draw this distinction, which is one reason the Tibetan commentarial tradition is richer on this topic.
異品, 異法, and 異法喻: The Class, the Rule, the Example
You will also meet two related terms: 異法 and 異法喻. They are not synonyms for 異品 — they are the verbal clothes 異品 wears when it enters the argument.
- 異品 is the *category* (a set of instances). - 異法 is the *formulation* of the rule: "wherever the predicate is absent, the reason is also absent." - 異法喻 is the *worked example*: "for instance, empty space (虛空) — it is permanent, and it is not produced."
The 《因明入正理論》 gives the canonical異法喻 as: 「若是常，見非所作，如虛空等」 — "Whatever is permanent is observed not to be produced, as in empty space." Here "常" (permanent) expresses the *absence* of impermanence (the property to be proved); "非所作" (not produced) expresses the absence of the reason. So the 異法喻 is doing double duty: it confirms both that the predicate is absent in this case, and that the reason is absent in this case — which is exactly what 異品遍無性 demands.
The relationship between 異品 (the class), 異法 (the rule), and 異法喻 (the example) is described in the tradition as 体用相即 — "substance and function coinhere." The class exists in order for the rule to operate; the rule operates by pointing at the example. Strip any one away and the others lose their grip.
The Domain Restriction You Cannot Violate
A subtle but crucial point, unique to the 《正理滴論》: 異品 is not "any random counter-instance you happen to think of." Its membership is constrained by the 因's own scope: 果因 (effect-reason) is judged on its fruit; 自性因 (property-reason) on its own nature. The text puts it sharply: 「於他處，不能任意論述同品及異品之有與無」 — "Outside that domain, one may not arbitrarily assert the presence or absence of 異品."
In modern terms: you cannot import counter-examples from a different comparison class and pretend they defeat the argument. If you are arguing "this drug treats depression by raising serotonin," you cannot defeat the argument by waving at, say, "but it also helps with anxiety!" — unless anxiety and depression are properly within the 因's intended domain. The 因 must be tested only against instances whose inclusion or exclusion is determined by the 因's own scope. This is a feature, not a bug — it prevents the argument from being trivially defeated by an irrelevant example.
How 異品 Failure Modes Look
When the third leg fails, the reason becomes an *indecisive* or *contradictory* reason (似因). The two classical traditions catalogue the failure modes in slightly different ways, but the core logic is the same. Per the 《因明入正理論》:
1. 同品一分轉，異品遍轉 — the reason is present in *some* similar instances but present in *all* dissimilar instances. E.g., arguing "sound is not effort-produced (勤勇無間所發) because it is impermanent" — but "impermanent" is found in dissimilar things (pots etc.) universally. Indecisive. 2. 異品一分轉，同品遍轉 — the reason is present in *all* similar instances but also in *some* dissimilar ones. E.g., arguing "sound is effort-produced because it is impermanent" — but impermanence is also in lightning (a dissimilar instance). Indecisive. 3. 俱品一分轉 — the reason is present in *some* of both classes. E.g., arguing "sound is permanent because it is non-tangible" — but non-tangibility appears in both permanent and impermanent things. Indecisive. 4. 相違因 — the reason appears *only* in the dissimilar class. E.g., arguing "sound is permanent because it is produced" — but "produced" exists precisely in things that are *not* permanent. This isn't just indecisive; it points in the wrong direction.
The 《正理滴論》 adds a particularly vivid case: trying to prove "X is not an omniscient one (一切智者)" by citing "X speaks." But 「言說與一切智二者不相違」 — "speaking is not incompatible with omniscience" (omniscient ones can speak too). So even using the dissimilar class to *remove* the predicate (遣除) doesn't work, because the 因 doesn't actually exclude the property. The reason is indecisive.
Across all of these, 異品 is the locus where the failure becomes visible. The first two legs might still be satisfied; it is only when you carefully examine the dissimilar class that you discover the reason is not as decisive as it seemed.
Walk-Through in Life
Most arguments we have — at work, in relationships, on social media — look like the structure of 因明 without our realizing it. We propose a claim (宗), offer a reason (因), and then never check the 異品 leg. Here are several contemporary scenes where it pays to walk the full structure.
Scene 1: The Dating Inference
Claim (宗): "He must love me." Reason (因): "Because he texts me every day." Similar class (同品): Other people who love someone and also text them every day — partners, devoted spouses, attentive family. "Texts every day" is present. ✓ Dissimilar class (異品): People who do *not* love me — friends, mentors, parents, colleagues. Do *they* sometimes text every day? Yes, often. ✗
The reason "texts every day" appears in 異品 too — it doesn't only show up where love is present. The third leg fails. 異品遍無性 is violated. The reason is indecisive (不定因). It might still be true that he loves you, but this reason does not establish it; you would need a property that is *absent* in all the non-love cases.
The common mapping mistake: treating 異品 as "people who obviously don't love me" (contradicting) while ignoring "people where the property is just absent in a quiet way" (彼上沒有) — friends, mentors. The 因 needs to be absent in all three sub-types from the 《正理滴論》, not just in the most obvious one.
Scene 2: The Productivity Self-Help Claim
Claim (宗): "Waking at 5am will make me successful." Reason (因): "Because successful people wake at 5am." 同品: Successful people who wake at 5am — present. ✓ 異品: Unsuccessful people. Do they sometimes wake at 5am? Of course — early-rising nurses, farmers, parents of infants, military recruits. ✗
異品遍無性 fails again. The reason is indecisive. Waking early may correlate with success in a way the 因明 tradition wouldn't deny, but the *causal* inference drawn from correlation is exactly the kind of conclusion that requires ruling out 異品 cases — and you cannot, because many non-successful people wake early too.
The common mapping mistake: treating 異品 as "lazy people who sleep until noon." That's the contradicting subtype (與彼相違之異品). But the broader surrounding class (包括彼外之異品 — every non-successful person, including those who wake at 5am for unrelated reasons) is where the 因 actually leaks. Confusing the narrow subtype for the whole class makes the argument look more decisive than it is.
Scene 3: The Medical / Research Inference
Claim (宗): "Compound X treats depression." Reason (因): "Because X raises serotonin." 同品: Other things that treat depression — do they all raise serotonin? Some yes, some no. 異品: Things that don't treat depression — exercise, certain foods, sunlight. Do they also raise serotonin? Yes. ✗
Again, 異品遍無性 is violated. The 因 is indecisise. The fact that X raises serotonin is not enough to conclude X treats depression, because the property "raises serotonin" is shared with things that don't treat depression.
The common mapping mistake: treating 異品 as "things that worsen depression" — the contradicting subtype. But the 因 leaks far more broadly than that: it leaks into the "彼上沒有" subtype, all the things where "treats depression" is simply absent but serotonin is still raised. The argument can only succeed if you find a property of X that is completely absent from every non-treatment.
This is exactly the structure of why drug development is hard, and why placebo-controlled trials are designed the way they are — the control group (the 異品, in essence) is the rigorous test of whether the 因 is decisive.
Scene 4: The Pattern-Recognition Inference (AI / Image Classifier)
A machine-learning team trains an image classifier to detect tumors.
Claim (宗): "This image shows a tumor." Reason (因): "Because the pixel pattern P is present." 同品: Other tumor images — pattern P appears in some. ✓ 異品: Non-tumor images — does pattern P appear here? If yes (some non-tumor images share that texture because of staining artifacts, scanner quirks, similar-looking tissue), the 因 leaks. ✗
Modern machine learning knows this problem intimately. It's why datasets need balanced negative examples (異品 cases); it's why classifiers overfit when given imbalanced training data; it's why fairness audits test models across demographic subgroups — each subgroup is in effect a 異品 test. The structure of the third leg of 因三相 is built into the design.
But — to flag a limit that the 因明 tradition would also insist on — 異品 is not "any negative image we can scrape off the internet." The 因's domain matters. If your model is meant for MRI scans, you cannot defend it against 異品 counter-examples drawn from CT scans or ultrasound; that would violate the 論域 restriction from the 《正理滴論》. The relevant 異品 is determined by the 因's own scope.
Scene 5: The Political / Media Inference
A headline: "Immigrant communities have higher crime rates; therefore, immigration reduces public safety."
Walking it through: - 宗: "Immigration reduces public safety." - 因: "Immigrant communities have higher crime rates." - 同品: Other things that reduce public safety — poverty, unemployment, lead exposure. Do they all correlate with high crime rates in immigrant communities? Mixed. - 異品: Other communities — do *non-immigrant* communities with similar socioeconomic profiles also show elevated crime rates? Frequently yes. ✗
異品遍無性 fails. The 因 (high crime rates in immigrant communities) appears in 異品 too — it shows up in non-immigrant communities that share the relevant background variables. So the 因 cannot be what's *causing* the conclusion; the 因 is being driven by a confounding variable that exists across 異品 as well.
The common mapping mistake here is treating 異品 as "law-abiding communities" only, ignoring the broader surrounding class. The 因 fails because of cases that aren't obviously "the opposite" — they're just cases where the property "reduces safety" is absent for reasons unrelated to immigration.
Why Contemporary People Need This
We live inside an information environment that is structurally designed to feed us 同品 without 異品. Social media algorithms surface content that confirms what we already lean toward; news headlines are framed to provoke a specific inference without testing its 異品 leg; productivity and lifestyle advice is built on anecdotal 同品 ("they did X and succeeded!") with the 異品 leg deliberately suppressed.
The 因明 tradition's insistence on 異品遍無性 is a sophisticated, ancient answer to a problem we now recognize by other names:
- Confirmation bias in psychology is, in structural terms, the habit of satisfying 同品定有性 while ignoring 異品遍無性. - Survivorship bias — drawing conclusions from winners without checking non-winners — is failing the 異品 leg. - Selection bias in data — drawing conclusions from available positive examples while missing the dissimilar class — is the same structural failure. - The discipline of falsification in modern scientific philosophy (associated with Karl Popper and others) is remarkably close to 異品遍無性 in spirit: a theory must be refutable; one well-placed counterexample defeats it. This is not a coincidence — both traditions are trying to keep inference honest by demanding the third leg be walked.
A useful analogy: think of a 因 (reason) as a key, 同品 as the locks it opens, and 異品 as the locks it must *fail* to open. If the key opens some locks but also opens a single lock in the dissimilar set, the key is not specifically for the locks in the first set — it is a more general key, and your inference is invalid. The whole discipline of 異品遍無性 is the discipline of demanding that the key be narrow enough to fit the 同品 locks and strictly fail the 異品 locks.
One limit to flag: 因明 is not "proven by" cognitive science or AI. The structural similarity is suggestive. We don't have a clean empirical demonstration that 因三相 optimizes reasoning. But the parallels are real, and contemporary logic, statistics, and machine learning have independently arrived at very similar structural constraints — which is itself worth noticing.
There is also a deeper payoff beyond argument hygiene. 異品遍無性 is a discipline of *generosity* — it forces you to actually consider the cases where your view doesn't apply, where the people you disagree with aren't villains, where the alternative explanation might also fit. Walking the third leg is, in practice, a small exercise in humility: a way of being less wrong about the world by refusing to ignore the cases that don't fit.
Common Misreadings and Clarifications
1. "異品 means 'opposites' or 'things against the claim.'"
No. 異品 is defined strictly by the absence of the property to be established (所立法), not by being the moral, political, or rhetorical opposite of the claim. For "sound is impermanent," the 異品 is things that are *not impermanent* — including things where impermanence is just absent, not actively contradicted. The 《正理滴論》's three subtypes — including "彼上沒有" (simply absent) — exist precisely to block this confusion.
2. "異品遍無性 means the reason is rare or uncommon."
No. "遍" means universally. Even one counterexample in 異品 collapses the third leg. The reason must be absent in *every* member of the dissimilar class within the 因's domain.
3. "異品 is whatever counterexamples I can think of."
No. The 《正理滴論》's 論域 restriction is explicit: 異品 membership is determined by the 因's own scope — the fruit of an effect-reason, or the nature of a property-reason. Importing counterexamples from outside that scope is illegitimate.
4. "異品 = 異法 = 異法喻. They're all the same thing."
Not quite. 異品 is the class (a set of instances). 異法 is the rule stated in words ("wherever the predicate is absent, the reason is also absent"). 異法喻 is the worked example ("as in empty space"). The 《因明入正理論》 uses these terms 互相發明 (mutually illuminating), but they have distinct roles: 異品 is the substance, 異法 is the formal structure, 異法喻 is the concrete instance.
5. "If the third leg fails, my conclusion must be false."
No. The third leg failing only means the specific reason you offered (因) does not establish your conclusion. Your conclusion might still be true; you just need a better reason. The 因明 tradition is rigorous about the relationship between 因, 宗, and 異品 — failing in the 因 does not entail failing in the 宗.
6. "異品 is about being negative or pessimistic."
No. Walking the 異品 leg is not defeatism; it's precision. The Buddhist logicians were not arguing that we should always expect failure; they were arguing that any honest inference must survive contact with the dissimilar class. A 因 that passes 異品遍無性 is a stronger, more reliable inference — and that's good news for anyone who cares about getting things right.
7. "The 因明 tradition invented a uniquely Buddhist logic."
Half-true. The formal apparatus of 因三相 was developed in the broader Indian *pramāṇa* tradition (where Nyāya was influential); Buddhist logicians like Dignāga and Dharmakīrti refined and reformulated it. So 異品 is a term with deep roots in classical Indian epistemology; the Buddhist usage is a careful, technical inheritance rather than a standalone invention. Naming this lineage honestly matters — 異品 didn't appear from nowhere.
Closing
異品 asks a quiet, unglamorous question: *in all the cases where the property you're trying to prove is absent, is your reason also absent — universally, not just usually?* It is the third leg of a three-legged stool whose other legs test positive presence (in the subject and in similar instances). Without it, any inference with two confirmed positive legs can still wobble, because the reason might be doing more work than the argument admits — pointing at a broader pattern, a confound, a coincidence, or a property other than the one named.
In a culture that constantly serves us 同品 (winners, success stories, exemplars, confirmations) and hides the 異品 (the cases where the same pattern shows up without our preferred conclusion), the discipline of asking the third-leg question is a small but sturdy piece of intellectual hygiene. It is also, in a deeper sense, a discipline of attention — a way of letting the world's inconvenient cases into the room before we settle on what we think is true.
The Buddhist logicians did not promise that walking the third leg would make us happy, or kind, or enlightened. They promised something smaller and more useful: that our reasons would *actually be reasons* — and that is a promise still worth holding them to.