Problem of Universals (སྤྱི་, sāmānya)
Do universals (general properties) exist in reality, or are they merely conceptual constructs? This is the question dreyfus-recognizing-reality is organised around, and the one that separates the Sakya from the Gelug reading of Dharmakīrti.
The Problem
We attribute common identities to things spread across space and time. We say “this is a lake” and “that is also a lake.” Three levels of commonality are at stake:
- Spatial spread: the unity of an object composed of many parts (the lake is made of water molecules)
- Temporal spread: the identity of the same object over time (the lake tomorrow is “the same” lake)
- Cross-object commonality: the shared property across different objects (lakeness across different lakes) — this is the universal proper
Positions
Nyāya Extreme Realism
(See nyaya and vaisesika for the schools and the full set of refutations.) Universals are fully real entities (kinds, རིགས་, jāti) that exist separately from their instances. Connected to instances by the relation of inherence (samavāya). Combined with the doctrine of substance (dravya) — commonsense objects are real wholes distinct from their parts.
Dharmakīrti’s Antirealism
Only individuals are real. All three types of commonality are conceptual fictions:
- There are no real wholes — commonsense objects are just convenient labels for aggregates
- There is no real persistence — duration is a fiction produced by the quick succession of similar moments
- There are no real universals — commonality across objects is a product of conceptual exclusion (apoha)
Sakya Antirealism (Sakya Paṇḍita, Gorampa, Sakya Chok-den)
Follows dharmakirti closely. Universals and individuations (བྱེ་བྲག་, viśeṣa) are entirely conceptual. The distinction between universals and individuations holds in thought only.
The Sakya curriculum draws on the སྡེ་བདུན་རབ་གསལ་ (Gorampa’s Clarification of the Seven Treatises) to fix two parallel lists, equivalent in extension within each (dzongsar-class-cognition):
| Real (functional) side | Conceptual side |
|---|---|
| specifically-characterised (རང་མཚན་) | generally-characterised (སྤྱི་མཚན་) |
| Instance (གསལ་བ་) | Universal (སྤྱི་) |
| Functional thing (དངོས་པོ་) | Other-elimination (གཞན་སེལ་) — see apoha |
| Substance (རྫས་) | Reverse (ལྡོག་པ་) / exclusion (རྣམ་བཅད་) |
| Ultimate (དོན་དམ་པ་) | The mixed (འདྲེས་པ་), relation (འབྲེལ་བ་), conventional (ཀུན་རྫོབ་) |
All on the left perform a function (དོན་བྱེད་ནུས་པ་) and abide unmixed in place, time, and nature. None on the right perform a function.
A standard Sakya argument shows that meaning-universals are themselves unmixed: each meaning-universal exists in one person’s body, at one time, with its own nature — so the realist cannot get the metaphysical “mixed” thing they need. Universals as the realist conceives them are not found anywhere.
Definition of particular (བྱེ་བྲག་, drawn from the Sde-bdun rab-gsal): a superimposition (སྒྲོ་བཏགས་) appearing as a combination of two reverses (ལྡོག་པ་ཉིས་ཚོགས་) — reversed both from the dissimilar class (རིགས་མི་མཐུན་) and from [some members] of the similar class (རིགས་མཐུན་). Hence: if X is a particular of Y, X is not necessarily Y. The meaning-universal of śapa is a particular relative to the universal “tree” but is not itself the universal “tree.”
Three-angle refutation of universal-instance same-substance (from the Sde-bdun rab-gsal, expanded in dzongsar-class-cognition):
- By entity (ངོ་བོ་): is the universal “tree” the substance of śapa? If yes, shukpa (juniper), being one substance with that universal, would be śapa (absurd). If no, they are not one substance.
- By cause (རྒྱུ་): did the universal arise from the cause of śapa? If yes, blocking the cause of śapa blocks the universal; if no, no causal connection — therefore not one substance.
- By arising and ceasing (སྐྱེ་འཇིག་): if the universal “tree” arose and ceased with each śapa, then every spatially, temporally, and aspectually distinct instance of “tree” would arise and cease in lockstep with this śapa — absurd.
A practical corollary: a vase is not established as a substance and is therefore unsuitable as a similar example for proving sound impermanent — “a permanent vase does not work as a similar example.”
gorampa defines universals as “a commonly appearing entity which is superimposed” and enumerates three divisions: kind universals (རིགས་སྤྱི་), meaning universals (དོན་སྤྱི་), and collection universals (ཚོགས་སྤྱི་). He notes that not only universals but also wholes and durations are non-entities (དངོས་མེད་). gyel-tsap presses the cost of that last item: deny any reality to duration and you undermine the understanding of karma and its fruits, since an action and its ripening have to be the career of something that lasts. The page records the objection without an answer to it (sonam-introduction-treasury).
In gorampa-pramanavartika (on PV 1.143–150) Gorampa systematically refutes the Nyāya-Vaiśeṣika real universal by a pair of dilemmas: a single universal within many particulars either abides dependently on those particulars (and then cannot be permanent, since dependence requires causal effect, which permanence excludes) or is illustrated by them (and then must stand in a cause-effect relation to the consciousness that apprehends it, which a permanent universal cannot do). Neither horn works — “how can that universal, which is to be illustrated, arise from those particulars, which contain the universal?” The positive account: what appears to conceptual cognition as a universal is a mistaken designation — “those phenomena that exist for consciousness are in this way possessed of a nature of being isolated; therefore they are as if not of distinct types, also appearing excluded from others, and that is a universal” (PV 1.78).
Gelug Moderate Realism (Gyel-tsap, Kay-drup, Ge-dün-drup)
A universal is “that phenomenon which is concomitant with its manifold individuals” (Pur-bu-jok’s Collected Topics definition). Universals have a moderate reality — they are not separate from their instances but exist as real properties dependent on them. The relation between universal and particular is “distinct distinguishers within the same entity” (ངོ་བོ་གཅིག་ལ་ལྡོག་པ་ཐ་དད་) — an attempt to distinguish without separating (dreyfus-recognizing-reality, pp. 171–172).
The moderate realist’s predicament
Dreyfus identifies a structural difficulty that recurs across Indian, Tibetan, and European philosophy: any moderate realism must specify a relation between universal and particular that is less than real (since they cannot exist apart) yet more than purely conceptual (since the realist insists they are not merely mental constructions). Duns Scotus calls this the “formal distinction”; Kumārila calls it “identity in difference” (bedhābedha); the Jain Samantabhadra makes universals and individuations relative aspects of substances; Gelug thinkers (Cha-ba, Gyel-tsap, Kay-drup) use “distinct distinguishers within the same entity.” Dreyfus argues that none of these moves is fully coherent — the moderate realist tries to occupy a position between two intelligible extremes and ends up with an intermediary that is hard to characterise without sliding back to one or the other (dreyfus-recognizing-reality, pp. 136, 139–40, 446–47).
Conceptualism, not nominalism
Dreyfus pushes back against the common reading of Dharmakīrti as a nominalist (which would mean universals are merely linguistic). He argues Dharmakīrti is a conceptualist in Quine’s sense: universals are mind-dependent constructs that exist in a weak sense as the objects/contents of concepts, not just as words. This distinguishes him from Sellars/Goodman-style nominalism on one side and from Locke/Berkeley-style abstractionism on the other: Dharmakīrti is a constructivist (concepts are actively built up through exclusion) rather than an abstractionist (concepts as abstracted features of given particulars) (dreyfus-recognizing-reality, pp. 133–34).
Distinguishers (ལྡོག་པ་, vyāvṛtti)
A key technical term in the debate. For Sakya antirealists, a distinguisher is a purely conceptual differentiation. For Gelug realists, distinguishers are real properties that things instantiate. sakya-chok-den criticises the Gelug view that “a thing instantiates its own distinguisher” as confused — collapsing the distinction between practical and critical understanding (dreyfus-recognizing-reality, pp. 167–168).
What the answer decides
The position taken here fixes three others, which is why the dispute runs through the whole tradition rather than staying in ontology.
- If universals are unreal, words cannot mean by naming them, and something else must do the work: apoha.
- If universals are unreal, inference operates on generally-characterised objects that are not part of reality, and its validity has to be secured some other way: can-inference-be-valid.
- If commonsense objects are conceptual syntheses rather than individuals, what perception delivers is not what we ordinarily take ourselves to see, which is the disagreement recorded below under Commonsense Objects.
Berzin’s Terminological Analysis: “Synthesis” over “Universal”
Berzin (berzin-science-of-mind, Syntheses, Categories and Individual Items) argues that no single English word — “universal,” “generality,” “category,” “synthesis” — adequately covers the range of meanings of the Tibetan spyi (and Sanskrit sāmānya). Different meanings require different renderings:
| Type | Tibetan | Best English | Status |
|---|---|---|---|
| Collection synthesis | ཚོགས་སྤྱི་ (tshogs-spyi) | “synthesis” (whole/parts) | Objective entity (Gelug) / static (non-Gelug) |
| Kind synthesis | རིགས་སྤྱི་ (rigs-spyi) | “synthesis” (type/instances) | Objective entity (Gelug) / static (non-Gelug) |
| Object/meaning category | དོན་སྤྱི་ (don-spyi) | “category” | Static (both traditions) |
| Audio category | སྒྲ་སྤྱི་ (sgra-spyi) | “category” (acoustic) | Static (both traditions) |
The term “universal” works only for static object/meaning categories; “synthesis” is better for collection and kind types. “Category” works for the static types. The same Tibetan syllable spyi also covers the concept of “metaphysical entity” (སྤྱི་མཚན་), which is why translations can be deeply misleading.
Commonsense Objects: The Battleground
What the dispute costs is clearest over commonsense objects — apples, dogs, tables, the things we take ourselves to see and handle:
- Gelug: commonsense objects are objective entities (rang-mtshan), cognizable by non-conceptual sensory cognition. They are collection syntheses (imputations on spatial/sensorial/temporal parts) and kind syntheses (instances of types), but these syntheses are themselves nonstatic objective entities.
- Non-Gelug: commonsense objects are metaphysical entities (spyi-mtshan), cognizable only by conceptual cognition that mentally constructs and labels them. Only moments of sensibilia exist as objective entities. Commonsense objects are not real in the sense of being individually characterised phenomena.
Both traditions agree that commonsense objects exist and operate causally (apples grow on trees, can be eaten, rot). The dispute is entirely about their ontological status and which type of cognition can apprehend them.
Sources
- dreyfus-recognizing-reality — the central thread of the entire book (Part II of Book One, Chapters 6–10)
- sonam-introduction-treasury — Gorampa’s threefold division of universals; Gyaltsab’s warning that denying duration undermines karma; perception as “complete engager” vs. conceptual cognition as mistaking objective aspects for reality
- berzin-science-of-mind — terminological analysis of spyi; three types of syntheses/categories; Gelug vs. non-Gelug on the status of commonsense objects
- gorampa-pramanavartika — Gorampa’s refutation of the real universal (PV 1.143–150); the many isolates of one entity (PV 1.40–42); universals, wholes, and durations as non-entities designated by mistaken conceptual cognition