CrossRef citations to date
Research Article

Ethics of atomism – Democritus, Vasubandhu, and the skepticism that wasn’t

Received 19 Dec 2022, Accepted 24 Jul 2023, Published online: 30 Oct 2023


Democritus’ atomism aims to respond to threats of Parmenidean monism. In so doing, it deploys a familiar epistemological distinction between what is known by the senses and what is known by the mind. This turns out to be a risky strategy, however, leading to inadvertent skepticism with only diffuse and contrary ethical implications. Vasubandhu’s more explicitly metaphysical atomism, by contrast, relies on a different principle to get to its results, and aims to address different concerns. It leaves us with a view that positively implies a concrete mode of practical engagement, and resources for a critical stance. Even if certain atoms end up proven incoherent, there is no danger of slipping into the morally fatal indifference of inadvertent skepticism. For the ethical implications, it matters how one arrives at one’s atomism.

Part 1: Democritus

Jonathan Barnes describes Democritus’ view as “a particulate theory of matter”,Footnote1 and there is no reason to dispute this characterization – in spite of the fact that Barnes then goes on to say, with equal justice, that “atomism begins with metaphysics”; and at the same time that “beings are plainly bodies”, and that with Democritus “we have a physical, not a metaphysical, hypothesis”. Atomism in Europe began, and remained, on the borderline between physics and metaphysics, and more explicitly connected to the natural sciences than to the moral ones.Footnote2

According to Democritus, the primary, fundamental, ultimate, most real (indeed the only really real) beings are indivisible bodies. These bodies are infinite in number,Footnote3 and come in various shapes and sizes and possibly weightsFootnote4 (mostly extremely smallFootnote5); they move, accumulate – change position and relation and orientation;Footnote6 but they do not themselves change. They are impassable, or solid;Footnote7 and being indivisible, they could not alter in Aristotle’s sense of it – what is atomos, indivisible, is partless, so it cannot have one part persisting over time and ‘underlying’ a change in some other part.

These atoms are in motion and, being impassible, “collide and become entangled” (Simplicius, de caelo 295.11 (KRS 583)), and thus “effect the coming into being of compound bodies”.Footnote8 Although these indivisible bodies “hit upon one another and bind together with a binding that makes them touch and be next to one another, [this] does not generate any genuinely single nature whatever out of them”; for “it is absolutely silly to think that two or more things could ever become one” (Simplicius, de caelo 295.11 (KRS 583, DK 68A37, tr. Barnes, emphasis mine)).Footnote9 There is no real unity composed of several things, nothing over and above the many constituents; just their respective weights, shapes, sizes, their relations and orientations, which give rise to appearances of macroscopic objects in us.

Democritus’ atomism picks up from his predecessor in atomism, Leucippus, an overriding concern to address the spectre of Parmenidean monism.Footnote10 Parmenides offered an analysis of being on which reality is one and indivisible, and motion and change are impossible – a position epitomized by Zeno in his paradoxes of motion. Democritus responds to this intolerable monist suggestion with an atomism that preserves the possibility of change (of a sort), while respecting principles of identity and being articulated by Parmenides.

On the atomist view, motion is possible by fiat: we declare that bodies are of finite, rather than infinitesimal extension, posit a void into which these finite bodies can move, and this resolves Zeno’s paradoxes.Footnote11 It also gives us a ‘real reality’ that can underwrite the phenomena (or reality as we experience it). Once motion is possible, then all apparent alteration is the changing place and orientation of the finite, indivisible bodies. Thus Democritus respects the Parmenidean intuition that one thing cannot be two,Footnote12 while demonstrating that this need not lead to absolute monism. There is indeed change in reality taken as a whole (and so Democritus can explain our varying experiences); yet the real beings themselves do not change – so in a way, Parmenides was right, what is real (any real thing) is unchanging.Footnote13

1.1. Democritus’ epistemology

Democritean atomism thus makes a strong and clear appearance-reality distinction, highlighted in the perhaps most oft-repeated idea associated with Democritus: “By opinion sweet, by opinion bitter, by opinion hot, by opinion cold, by opinion colour; but in reality atoms and void” (Sextus, adv. Math VII.136, DK68B9, my trans.).Footnote14

The word nomōi here is usually translated as ‘by convention’, picking up the meaning of the noun, nomos, of ‘custom’. But Democritus cannot mean this in the sense of ‘consensus’, since his point is precisely the non-consensus on perceptible qualities;Footnote15 and there is evidence that he himself glossed nomōi in this context with nomisti, perhaps picking up the verbal root nomizein – to think, judge or opine.Footnote16 Indeed Sextus introduces the Democritean dictum by describing sensations as being ‘kata doxan’ (according to judgement/opinion) as opposed to ‘kata alētheian’ (according to truth), and his elaboration of the meaning of nomōi elsewhere is not by reference to social conventions, but rather “sensibles are considered [νομίζϵται] and judged [δοξάζϵται] to exist” (adv. Math VII.135). So the emphasis in the contrast is on what is merely thought, believed or judged – or indeed exists only by being so judged – as opposed to what really is.

Thus, as the connotations of ‘cognition’ in nomōi as well as the selection of things that are nomōi suggest, Democritus will make his appearance-reality distinction epistemologically, according to the route by which our cognitions come to us. What is merely considered so is whatever comes to us through a dubious route; whatever is in fact so is what comes to us through a more reliable source. Sextus’ formulation captures this epistemological distinction driving the metaphysical view.

In his Canons he [Democritus] says there are two kinds of knowledge [δύο … γνώσϵις], that which is through the senses [διὰ τῶν αἰσθήσϵων], and that which is through the mind [διὰ τῆς διανοίας]. Of these, he calls the one through the mind ‘legitimate’ [γνησίην], ascribing to it credibility [πιστὸν] for discriminating the truth [ϵἰς αληθϵίας κρίσιν], while the one through the senses he names ‘bastard’ [σκοτίην], depriving it of infallibility for discernment of truth [ἀφαιρούμϵνος αὐτῆς τὸ πρὸς διάγνωσιν τοῦ ἀληθοὺς ἀπλανές].

(adv. Math. VII.138, my trans.)
Sextus specifies the bastard sort of cognition further in a direct quotation of Democritus, “Of knowledge there are two forms, the genuine and the bastard: and to the bastard belong all these – sight, hearing, smell, taste, touch; but the other form is distinct from this and genuine”.Footnote17 In short, sensible properties – what comes to us via sense-perception – are dubious, merely how things are considered to be; what is known through the mind is really so.

This distinction between intellect and senses is familiar in the European tradition; Plato adopts and develops it extensively.Footnote18 It is so familiar as to seem unproblematic, and yet the problems it raises are themselves a familiar part of the European tradition. Of the two truth-apt experiences available to us, sensation (αἴσθησις), as opposed to reasoning or intellect, is unreliable and idiosyncratic. It gives us awareness only of our own bodily changes, how we are affected by things; it can tell us nothing of the nature of things themselves (Sextus adv. Math. VII.137 glosses it this way). Theophrastus plausibly suggests (de Sens 63–64; DK 68A135) that Protagorean-type concerns underlie Democritus’ suspicion of sensation: Sensible particulars, for Democritus, are not in reality because different creatures experience things differently.

Thus, on Democritus’ view no less than on Parmenides’, the world is not as we experience it. There are no middle-sized dry goods – it took Aristotle to try to salvage that phenomenon; there are only unchanging, indivisibles of varying shapes and sizes, moving in space so that position and orientation and thereby relations to one another vary over time, thus creating the appearance for us as of middle-sized dry goods, and as of living organisms growing and reproducing. All sensible properties are not properties of these bodies, but are rather ways that we are affected by the size and shape, motion, position, and relations of these indivisible bodies.

1.2. Democritus’ ethics

If there were to be an ethics associated with this form of atomism, it would likely take a Platonic stripe, in virtue of this fundamental mistrust of the senses as authoritative in preference to reason.Footnote19 “Faring well and poorly belongs to the soul” (DK 68B170-71 = Stob. 2, 7, 3i), and it is consistently ascribed to wisdom, intelligence or calculation (e.g. DK 68B197; see also DK68B2, B160), while all sorts of vice is ascribed to and described as failures of intelligence (e.g. DK 68B160, 199–201). And indeed, among the (copious and anodyne) ethical fragments of Democritus, we find, for instance

Good and true [ἀγαθὸν καὶ ἀληθές] are the same for all men; but pleasures differ for different persons.

[DK 68B246]

You must not choose every pleasure, but only that taken in what is good.Footnote20

While DK 68B83 tells us, “The cause of mistake is ignorance of what is better” (ἀμαρτίης αἰτίη ἡ ἀμαθίη τοῦ κρέσσονος), DK 68B235 invokes an independent criterion of appropriateness, and DK 68B191 appeals to measure and symmetry. As Kirk, Raven, and Schofield wryly observe, some of the content of Democritus’ ethical fragments “seems embarrassingly close to Socratic or later doctrines” (Presocratic Philosophers, 431).Footnote21

Of course, this is very much not the direction in which Epicurus took Democritus’ thought, and there are certainly materials enough in the disparate fragments for the ethical direction Epicurus did take.Footnote22 But then notice crucially that Epicurus inverts Democritus’ epistemology, too, legitimizing the offspring of sense-experience at the expense of reason, and thus validating an affective euthymia over other aspects of Democritus’ ethical thought.

While Democritus’ ethics is far from univocal, the Platonising strand within it is, I think, no accident, but rather is closely related to Democritus’ epistemic way of making the appearance-reality distinction, according to our mode of access. For it is by picking up this distinction that Plato infamously ends up committed to a real reality not at all atomist. He sees – as perhaps Democritus did not – that once the senses are called into question, no physical properties are safe. Weight, solidity (impassivity), shape, and size are no more immune from whatever infection ‘being known via the senses’ carries than are colour or heat. But it was in no way Democritus’ intention to go the way of Platonism – his physicalist atomism was meant to save natural change from Parmenidean onslaught, not to redeem a whole world of changeless, non-sensible Parmenidean ‘ones’.

There is a related difficulty. The more properties that fall under the dubious, sensible sort, the more obscure the supposedly pure sort of cognition becomes. The only pure cognition will be that which operates entirely independently of experience. And whatever that tells us – if anything at all – it will tell us nothing about, well, experience. The impure cognition, for its part, tells us only about experiences, and not the things experienced. So any access to ordinary things – cats and cows and carbuncles – seems ruled out entirely: they cannot be known, because they do not really exist (only atoms and void really exist); nor can they be perceived by sense-perception, for that is only of our affects, and a cat is not an affect.Footnote23

Quite likely Democritus anticipates something of this, for he is recorded fairly early on as having a positively dismal view of our possibilities for attaining knowledge at all. “Which of them [the sense-impressions] are true or false is unclear, for these are no more true than those, but of equal standing”, writes Aristotle, in Metaphysics Γ5 (1009b7). “This is why Democritus, at any rate, says that either nothing is true, or it is unclear to us”. The difficulty seems to be that intelligence has no canon or criterion for adjudicating one sense-impression reliable and another not. The only resources intelligence could appeal to would be those provided by sensation itself – and these are deemed congenitally unreliable. And yet, “Wretched mind, do you take your assurances from us and then overthrow us? Our overthrow is your downfall” (KRS 552).Footnote24 It is hardly any wonder, then, that Democritus was also taken by some in antiquity for a skeptic.Footnote25 Diogenes Laertius tells us Pyrrho thought highly of him; and apparently his own student, Metrodorus, became a skeptic.

Small wonder too that Sextus Empiricus later cheerfully presses Democritus into service in the skeptic cause.Footnote26 But as he also drily observes, no one who says “in reality atoms and void” – or “in reality” anything at all – can be a skeptic (M. 1, OP I.xxx). But since it is hard to see what could warrant the assertion of atomism except the need to explain the appearance of change – that is, experienced phenomena – the precipice of skepticism is near at hand.Footnote27 I call skepticism a ‘precipice’ in Democritus’ case because there is no hint that he locates the source of our unhappiness in our indecision or all-too-dogmatic judgements, nor that he thinks we would be either better people or happier if we refrained from judgement altogether. If Democritus’ atomism leads to skepticism he has wandered into it accidentally, and an accidental skepticism is a dangerous thing.

1.3. Summary of Democritus

In sum, Democritus distinguishes mere appearance from real reality epistemologically, according to our route of cognitive access to each. But this is a dangerous way of proceeding because, as seen above, there is pressure towards putting more and more into that which is arrived at through deviant, impure means, leaving nothing to be the object of the pure means of cognition, by which we cognize reality instead of our appearances – and no way of bringing that reality to bear on the adjudication of conflicting appearances. The pressure increases to ask Descartes’ question – whether this supposedly kosher form of cognition itself is not just another way appearances arise for us – and to find no satisfactory reply. Not only are weight, size and shape, on examination, just further deliverances of the senses, but also conversely all manner of non-sensory cognitions qualify as mere appearances. It appears to us that there are horses and houses, but in reality there are only atoms and void. Such conceptually rich appearings, as Plato argues in the Theaetetus against ‘Protagoras’, are just as non-adjudicable as diverse sense-impressions; and they are not limited to appearances of objects but extend also to appearances of goodness and benefit.

These appearances themselves are, as Aristotle says and Protagoras said before him, all equally valid and true, for there is no criterion of validity other than their appearing. If what is available to rational cognition is vanishingly little and in any case utterly cut off from everyday experiences, which are one and all mere appearings and ‘by opinion’, all on a par, then even the doctor’s expertise is nothing other than the power to induce impressions of wellness in others. By rejecting the Protagorean claim that such appearings are knowledge, you may get to keep your epistemic humility, but it is an abject humility; for it comes at the price of all appearings being equal, with no reasons to count anything better or worse, nor closer or further from a correct grasp of how things are. It is not just your atomism which has now flown out the window. As Plato emphatically underlined in the discussion staged between Socrates and Protagoras in the Theaetetus, where there is no reason – and no prospect for reason – there is only power.

Part 2: Vasubandhu’s atomism

Such inadvertent skepticism may not be a logical implication of Democritean atomism; but it is a natural trajectory from Democritus’ starting points. Atomism, however, need not unfold naturally in this direction. As an examination of Vasubandhu’s Abhidharma thought will show, a different set of motivations and arguments for (a different sort of) atomism, may lead naturally in quite different ethical directions. Vasubandhu’s atomism, I shall argue, is different in being fundamentally metaphysical, and not in the borderlands between metaphysics and physics. It is not ‘a particulate theory of matter’, and it has nothing to do with adherence to any presumption that ‘beings are plainly bodies’. This truly metaphysical atomism is arrived at through a particular application of the ‘one cannot be many’ principle that we saw evidence of in Democritus, above. While later Buddhist philosophers avail themselves freely of the abstract principle,Footnote28 earlier Buddhist thought sees the notion arising implicitly in the context of the self.

2.1. Origins of Vasubandhu’s atomism

Vasubandhu was active in the 4th century a.d., and his Treasury of Abhidharma, with Commentary (Abhidharmakośa-bhāṣya) offers a systematic exposition of one of the most prominent strands of Abhidharma philosophy, the [Mūla]Sarvāstivāda.Footnote29 Abhidharma philosophy itself is multi-vocal, comprising competing positions on a range of metaphysical, ethical, psychological, cosmological, and epistemological topics growing out of the first systemisations of the Buddha’s thought.

Among the earliest articulations of this thought from which Abhidharma philosophy drew, the signature ‘no-self’, anattā (Pāli; Sanskrit, anātman), claim is the most salient for our current examination of Vasubandhu’s metaphysics. An indicative passage from the suttas (Skt. sūtras) – the discourses of the Buddha recorded after his passing away – is this passage from the Connected Discourses:

If there is the view, ‘The soul and the body are the same’, there is no living of the holy life; and if there is the view, ‘The soul is one thing, the body is another’, there is no living of the holy life. Without veering towards either of these extremes, the Tathāgata teaches the Dhamma by the middle: ‘With birth as a condition, aging-and-death [arise]’.Footnote30

(SN II.61)
Other passages advise that “any kind of material form [or psychological moment] whatever … should be seen as it actually is with proper wisdom thus: ‘This is not mine, this I am not, this is not my self’” (MN 22, tr. Ñāṇamoli and Bodhi).

The metaphysical picture implicit in these claims is articulated by the nun Vajirā in a well-known image found in a passage from the Connected Discourses. Approached by Māra with the challenge, “By whom has this being been created? … Where has the being arisen? Where does the being cease?” Vajirā replies:

Why now do you assume ‘a being’? … This is a heap of sheer formations [Pā saṅkhāra]. Here no being is found. Just as, with an assemblage of parts, the word ‘chariot’ is used, so when the aggregates exist, there is the convention ‘a being’.

(SN 5.10, PTS I.135)

Notice how this sounds like Democritus in the distinction between reality (a heap of sheer formations is all that is really there) and convention. But then notice too how very different are the items picked out as belonging to each kind. Vajirā identified complex wholes – chariots and persons – as ‘by convention’; Democritus would likely agree about the chariots,Footnote31 but seems to have thought persons – the primary target of Buddhist deconstruction – were real.Footnote32 Conversely, the ‘hot’ and ‘bitter’ that Democritus singles out as dubious mere appearances would be among the ‘formations’ Vajirā counts as really existing without implying a self or ‘being’ as the bearer of these properties or haver of these experiences.

Vajirā’s aggregates – which, when heaped together, are called ‘chariot’ or ‘person’ or being of some sort – are all the particular and varied qualities, mental as well as physical, that arise and fall away. This particular volition is as real as this particular occurrence of red, or that instance of solidity. There is nothing especially suspect in the affects, or any other sort of mental event. Suspicion lies rather on apparent complex unities as such: the supposed one thing which all these many things comprise is what is dismissed as not really existing.

One of the reasons for this difference in what counts as really real is that the Buddhist does not rely on an epistemic distinction to distinguish between appearance and reality.Footnote33 Vajirā does not rest her distinction on a supposed difference in our cognitive powers, and she makes no tacit appeal to bastard and pure-bred forms of cognizing. Another key difference is that there is no threat of changeless monism standing in the background of any of these texts. That there is change is not in doubt, nor is it seen as being in need of justification (if anything, it was Buddhist flux theory that gave impetus to Vedanta monism, rather than the other way around).

Of course it might be churlishly observed that Vajirā does not rest her claim on a difference in our cognitive powers because she does not offer any argument at all – she just makes the claim. So let us turn to a much discussed passage from the Questions of King Milinda (the Milindapañha, composed probably some time between the first century before and the first century after the common era), which makes of Vajirā’s assertion a distinctively mereological argument.Footnote34

Taking up the chariot as analogous to persons, the monk Nāgasena explains to King Milinda that the chariot is neither identical to any one of its parts, nor identical to all of its parts, nor is it something quite separate from its parts. In this sense, there is no chariot; and in just this sense there is no Nāgasena or Milinda, either. Nāgasena is neither distinct from his psycho-physical constituents, nor identical to any one or several of them.Footnote35 Nevertheless, this does not make Nāgasena a liar when he gives his name, nor the king wrong when he says he came by chariot. For because of their respective parts, ‘chariot’ and ‘Nāgasena’ and all organisms or artefacts “exist as a denotation, appellation, designation, as a current usage, merely as a name” (MP I.28, tr. Horner). From this, we get the distinction between existing as a designation and really existing or existing ultimately (paramārthasat), which for the moment is simply ‘not like that’.Footnote36

Without analysing this argument for its success, it is clear that if it works at all, it works at every level of analysis, regarding any sort of complex whole, and pretty transparently so. If I can run the argument on chariots, I can run it on parts of the chariot; if it applies to the chariot’s wheel, then it applies equally to the wheel’s parts, the spoke and rim; and if to spoke, then to its parts. So similarly, if I am not my body, nor my mind, then neither is my body the flesh or the finger and neither is my mind this or that faculty or constituent cognition … and so on until there is no plurality of distinct constituents identifiable. The principle acting as the universal solvent in all this is:

Nāgasena’s Principle Anything partite is several things AND not one.

Quite generally, one thing cannot be several things; a complex thing cannot be one of its constituents; nothing complex exists separately from its constituents. Since one thing cannot be many, any cognition of a multitude as a single thing owes the unity cognized to the cognizing and not to the object cognized. Therefore all terms denoting a multitude as a single thing are mere designations; all cognitions as of unified multiplicities are ways of conceiving, and not recognitions of reality as it is. It is worth reiterating that this general principle is quite unrestricted. Since there is nothing here about extension and mathematical divisibility, there is nothing that restricts the argument’s validity to bodies. It is thus a short hop indeed from giving any reason at all for non-fundamentality of persons to a distinctively Abhidharma atomism.Footnote37

We have also along the way picked up the Abhidharma correlate to Democritus’ appearance/reality distinction in the difference Nāgasena draws between the constituents which really exist (exist ultimately) and chariots and persons and all other complexes which exist ‘as a designation’.Footnote38

2.2. Vasubandhu’s atomism and the principle of analysability

In his canonical codification of this Abhidharma distinction between existing ultimately and a secondary sort of existing, Vasubandhu thus draws the line rather differently from Democritus:

The idea of a jug ends when the jug is broken; the idea of water ends when, in the mind, one analyses the water. The jug and water, and everything like that, exist conventionally (saṃvṛtisat). Anything else exists ultimately.Footnote39

(AKBh VI.4)
‘Exists conventionally’ (saṃvṛtisat) is what has become of Nāgasena’s ‘exists as a designation, a description, an appellation, a current usage, a name’, informed moreover by the equivalence of samvṛtisat and prajñaptisat (conceptually real) in Abhidharma discourse.Footnote40 Vasubandhu then offers the following illustrative example:

If we grasp and remember the dharmas, such as colour, etc., in the water, then the idea of water will disappear. These things – jug, clothes, etc., water, fire, etc., – are given their different names from the conventional point of view, or in conformity with conventional usage.

(AKBh. VI.4)

Vasubandhu specifies two kinds of divisibility, physical and mental; to neither of these may what is ultimately real be liable. But since physical divisibility implies conceptual divisibility, but not the other way round, the latter is the more aggressive solvent, and the one really at issue. That is to say, complex things – whether physically separable or not, indeed whether they are physical existents or not – whatever is liable to conceptual analysis of any kind does not exist fundamentally or ultimately.

Such conceptual analysability is importantly different from what Barnes characterizes as “theoretically divisible”, when he presents Democritus as concerned about whether a body is divisible in principle, even if not in fact. For Vasubandhu, no worries about the infinite divisibility of extension stand in the background. Vasubandhu’s analysability is concerned rather with distinguishability – and concerned with whether the analysand retains its identity once distinguishable aspects, which are not identical with the analysand, are clearly identified as distinct, and “thought away”. Water may be both fluid and cool; but coolness is not the same as fluidity, and neither of them is identical to water. Yet, separate them out, and there is nothing remaining which is the water itself. This is the chariot argument at the microscopic level. And, having nothing to do with extension, it may operate as effectively in analysing mental existents, should there in fact be, as the Ābhidharmikas supposed, non-physical existents.

By contrast, form (rūpa: materiality, or non-mentality) is basic, for when you strip away any of the particular characteristics, the very notion of form itself, simple and unqualified, remains.

For example, rūpa: one can reduce rūpa into atoms [paramāṇu], one can remember smell and other dharmas in the mind, but the idea or the unique nature of rūpa persists. The same holds for sensations, etc. And this exists ultimately, is ultimately true.

(AKBh. VI.4)
Vasubandhu’s analysability criterion for ultimate reality resolves whatever is fundamentally real into absolute simples – and it does so without relying on prior conceptions of ‘whole’ or ‘part’.Footnote41 Substance-property metaphysics itself falls by the wayside, for whatever property could be distinguished from the substance it supposedly belongs to – even just mentally or conceptually distinguished from it – is simply a distinct thing, and the two together do not form a real unity. This position has therefore been likened to trope theory.Footnote42 But if it is a trope theory, it is a particularly sparse or austere one – by no means every predicate gets to be a trope (or a dharma). Nor is it sparse in a naturalistic way, as if we should wait on technological developments to determine which things happen to be indivisible. Rather, it is sparse for conceptual reasons: Nothing complex is ultimately real (two things cannot be one thing), so even shape and size, the primary properties of Democritus’ atoms, will not be ultimately real (ABKh. IV.2b-3b), for they involve a complexity which dissolves upon analysis of their distinct constituents.

Indeed even motion itself is only apparent and not ultimately real. For, Vasubandhu argues, true atoms cannot endure. The simples which remain once constituents are distinguished and thus “analysed away” must necessarily have only momentary existence (AKBh. IV.2b-3b); and a momentary existence cannot undergo change of place any more than it can undergo any other change. Vasubandhu’s ingenious argument for this would take us too far afield to consider here;Footnote43 but his position and his argument for it underscores the extreme simplicity of whatever may lay claim to being ultimately real.Footnote44

Thus, rather than Democritus’ eternally moving atoms, Vasubandhu offers just the arising and ceasing of property-particular events, or the instantaneous occurrences of simple properties, including mental property-events. Where Democritus preserved the appearance of change by making motion possible, Vasubandhu gives us just the appearance of motion through the actual occurrence of manifold simple and momentary events.

Part 3: atomism without skepticism

From Vasubandhu’s Analysability criterion for distinguishing conventional from ultimate reality, we might extract what we can call Vasubandhu’s Principle:

Whatever can be divided by mind is proved thereby to be put together by mind.

This understanding of the implications of VI.4 is supported by a comment Vasubandhu makes in another text, the Twenty Verses, with Commentary:

If there is a distinction of characteristics (lakṣana), then substantial nature (dravyāntaratva) is due to construction (kalpyate), and not otherwise.Footnote45

(Twenty Verses, Commentary ad v. 15, my translation)
This principle is based on the one-cannot-be-many principle, and implies a particular way of understanding the apparently complex: namely, as owing its complex identity to mental activity. Instead of Democritus’ bastard and pure-bred cognition, cognitions are divided by their contents according to whether these are simple or complex.

Now Democritus of course thought that the mere appearances were how things were thought to be, as opposed to how they actually are. But Democritus takes these mere appearances to be the deliverances of essentially passive sense-faculties being acted upon by a distinct portion of material-atoms reality. It is an inbuilt fault with the mechanism that the impressions could never convey the nature of the thing causing the impression. Vasubandhu by contrast, following faithfully the spirit of his predecessors, is much more attentive to the activity of our considering or taking something to be so. For Vasubandhu, appearances come from the mind illicitly making a one out of many. To say ‘there is the convention of saying’ is, for Vasubandhu, to acknowledge our active participation in the construction and construal of reality, rather than to point out the inevitably faulty deliverances of our essentially passive sense-faculties.

For Vasubandhu, the coming-to-be of anything ultimately real may be fully accounted for by other ultimately real things. Its identity and distinctness, however, its individuation as this, is due to itself and to no other source.Footnote46 What is complex, by contrast, inherits its existence from its simple constituents, while its identity as some one thing has a different source – namely, our taking the many as one. The identity of the chariot as a chariot is, in a way, real (the king is not, after all, a liar); but it is so due to ways that we collect and individuate simples. The reality of complex things is partially constituted by our practices of individuation (“These things – jug, clothes, etc., water, fire, etc., – are given their different names from the conventional perspective and conventional usage”).

In contrast to Democritus, Vasubandhu is not concerned with the sources of our information – a distinction that, as Democritus makes it, takes for granted the mind/body contrast and the integrity of the body and soul;Footnote47 Vasubandhu is concerned with the sources of a thing’s identity: Is that identity wholly subsisting in the thing itself, or is it constructed by mental activity? If the latter, then its reality is ‘conventional’, ‘conceptual’ and constructed.

3.1 Ethical implications of Vasubandhu's atomism

Identifying items as constructed is pertinent because constructing is something we do – not consciously, not deliberately, and certainly not in a way that we could just choose to do otherwise should we wish. Nevertheless, individuating one cluster of properties from another is something we contribute to our experienced reality. And wherever we do something there is some purpose. By highlighting the individuation of complex wholes as an active practice we engage in, such individuation is thus highlighted as motivated. Each of these points make all the difference for the ethical implications of the view. Vasubandhu’s atomism does not just avoid inadvertent skepticism; it thereby offers an angle for evaluation and critique of these ordinary experiences.

Ordinary experiences of complex wholes are not, as Democritus’ mere appearances, just hopelessly false and illusory. Vasubandhu’s way of distinguishing real from apparent tells us more: in particular, it tells us that complex wholes are put together by mental activity. Among the simples are some pleasures and pains, some inclinations and aversions, which vary according to how we individuate phenomena, and which in turn drive that activity of individuating.

Recall Democritus’ difficulty. When Democritus said “by opinion cold, by opinion hot”, he was taken as asserting merely that these properties are not really real – acknowledging the truth in Protagoreanism (these are just our impressions), while reserving some firm ground of real reality, however inaccessible to us. That these experiences of sensible properties could neither be interrogated nor reveal anything about anything except that we have such-and-such an experience can lead only to a pessimistic shrug at the dim prospects for true knowledge – a resigned and reluctant skepticism.Footnote48

It is resigned because there is nothing we can do to approach real reality, or adjudicate conflicting appearances. Every possible experience for us is equally removed from how things really are, and nothing we can do could even improve our cognitive situation. It is reluctant because Democritus wanted to save the appearances (namely, of change) and save a real and shared reality distinct from appearances, so that we do not get lost in a miserable Protagorean solipsism, or a lofty Parmenidean one. But if sense-perception is illegitimate cognition in the way he sets out, then his atomist theory does not even matter – it is just the way things appear to be to him, and it is not needed to explain ‘real’ change of bodies in space, because we cannot even know if there are any such things and even if there are, our only access to them is the appearances, which lie.

Vasubandhu’s atomism avoids these pitfalls, and moreover suggests a concrete alternative. Consider first the difference between Democritus’ epistemological basis for distinguishing misleading appearances, and Vasubandhu’s direct appeal to the simplicity or otherwise of the objects of cognition. Naturally, Vasubandhu’s principle, and the distinction between conventional and ultimate reality as the Buddhists make it generally, also recommends suspicion – suspicion not of sense-perception rather than intellect however, but suspicion of all of our experiences as of independent, complex wholes. In fact, Democritus’ bastard sensible properties are, for the Buddhist, one of the few things that could lay claim to not being conventional – if we but saw them truly. But even this suspicion of our cognitions of non-simples does not warrant their blunt dismissal for failure to be knowledge of real reality. It rather invites analysis into sources of such constructions, the motivations for them, and even evaluation of the purposes they serve.

To observe that ‘Nāgasena’ exists as a conventional designation does not mean that anyone might call him whatever they like, or reject the appellation altogether. But it focuses attention to the constructed nature of the shared practice of picking out that shifting and imprecisely circumscribed subset of simples as a single thing, for which it is useful to have a single name – and, indeed, the same name for us all, Nāgasena. To recognize its constructed or conventional nature is to see the usefulness of having such terms; and to acknowledge its usefulness is to be bound by the conditions under which it is useful – namely, in the first instance, that the practice be a shared one.

This presents already one advantage over Democritean atomism. For to recognize this as our activity is to be able to correctly locate the standards of correctness to which we are liable. In contrast to the hopelessly idiosyncratic nature of Democritus’ appearances, Buddhist conventional reality is a shared one, and there are acknowledged conditions for applicability of our conventional terms, as well as a now explicable degree of elasticity in these conditions. Being both shared and constructed in all cases for the purpose of communicating and coordinating with others, conventional reality can be known, claims can be adjudicated true or false, valid or invalid, by reference to a public standard of whether taking this subset of simples as a unity enables one to communicate and coordinate effectively.

So for Vasubandhu there can be knowledge and discernment among appearances – it is, after all, called conventional truth or reality. Names and concepts designate ways of bundling and individuating simples which enable us to conceive and achieve shared and unshared goals. This gives us standards of correctness, even where correctly discerning ultimate reality fails: calling Nāgasena ‘chariot’ is just false, even if neither Nāgasena nor the chariot exist as real individuals; and bundling together Nāgasena’s hair with the chariot’s wheel into a supposed single entity is not a creative extension of language and practice because such a synthetic ‘unity’ does not serve the purpose just as well. So even among what is not ultimately real, there is no danger of a flat Protagorean relativism.

3.2. Deconstructing and reconstructing reality

But we can go further. Vasubandhu’s non-epistemically based atomism retains not only a criterion of correctness in conventional truth, but also retains the prospect of coming to know real reality – of discerning the simples unconstructed, as they are.Footnote49 While we certainly do not ordinarily directly experience unconstructed simples, since there is no essential flaw in our cognitive capacities, there is no reason why we may not hope for improvement in this direction and perhaps even, ultimately, success in cognizing reality as it is. This has several implications, and will be the key to Vasubandhu forestalling the worry that his view escapes a simple personal relativism only to land in a sophisticated cultural relativism where our very agreement secures us against any critique.Footnote50

Most important from a Buddhist point of view, is simply that ultimate, real reality remains knowable, for only seeing reality as it is will truly eliminate suffering.Footnote51 In addition, recognizing the possibility for such knowledge, even before fully attaining it, involves the basic recognition of our constructed cognitions as constructed. This is already a better appreciation of how the world really is, and it has psychological as well as logical and practical implications. Psychologically, the pursuit of knowledge of ultimate reality so conceived engenders a certain wholesome detachment from the presumptive authority of conventional wisdom, and even from our own unquestioned aims and motivations. Conceptually, it contains implicit levers for modifying conventional reality as well as a stable cognitive ideal outside the claims of everyday life, by which to critically evaluate conventions.Footnote52

To recognize conventional reality as conventional implies a more accurate understanding of complex wholes as constructed, not given by nature. It also, in its implicit appeal to usefulness in adjudicating conventional truth, highlights the teleological character of the practices constructing our conventions: ‘useful’ is only what is useful for some purpose or another. Only because we aim to communicate and coordinate is it useful to have shared forms of individuating at all; and only because our specific goals and purposes are what they are do some, but not other, forms of bundling simples together prove efficacious. This means that to see ordinary experience as conventionally real at the same time first brings into view the role of aims and purposes in structuring our reality, which in turn points to the space where practical alternatives might get a foothold.

Moreover, coming to see real reality is not just a matter of drawing this general conclusion. Rather, the activity of knowledge-seeking involves indentifying in detail the particular elements, including the purposes and desires, that go into constructing complex wholes.Footnote53 This brings to light not just the degree to which our presumed categories and aims are dependent for their meaning and intelligibility on each other, our histories, our concepts for construing things and so on. It highlights precisely which elements depend on which in this complex system of dependencies, and habituates us into a practice of identifying these in our everyday experience. Identifying these dependencies is an opportunity– for seeing that prevailing categories structuring experience are not forced upon us gives us also the necessary tools to consider whether there might be other ways of individuating and evaluating that are preferable, and how these might be engendered.

If our ordinary reality is constructed so as to enable us to achieve our purposes and satisfy our aims, then this immediately suggests a point of internal critique ready to hand. For once granted that we carve the world so as to achieve our goals, we can always ask, ‘Does our current way of carving up and conceptualising experience in fact enable us to achieve our goals?’ Take, for instance, practices of punishment and justified anger – a feature of ordinary life to which Buddhist themselves devoted attention.Footnote54 We construct standards of guilt, notions of action, responsibility, blame and punishment for certain purposes. What are those purposes? And is the way we are doing it in fact achieving those purposes?Footnote55 Since we are well practiced in identifying the elements of experience that do in fact follow on from one another or are associated with one another, we are in a position to make a realistic assessment of which sorts of modifications in how we delineate multitudinous reality will effect changes which better meet our goals. And since which elements do in fact associate with or arise subsequent to one another is a matter of ultimate reality – and not of our preferences – we have a firm basis for debating the matter among ourselves.

But there is also the opportunity for more radical critique. If our distinctions and categories are not forced on us by an impervious reality, then neither are our goals or purposes. Once we have identified to what purpose, to what end, we are distinguishing things in this way, it is possible to raise the further question, ‘Should this be our goal? Are these intentions – and framing the world in terms of these intentions – actually doing any good?’Footnote56 Indeed, exactly this was the radical challenge the Buddha posed: Fame, pleasure, wealth, repute – all the supposed goods that move us to structure our more proximate ends – are fundamentally misguided.

Here the in principle possibility of real knowledge of ultimate reality becomes important. For in order to interrogate the worthwhileness of goals and purposes themselves, one must be able to appeal to something else more certainly good, by reference to which one purpose shows up more poorly than another. For the Buddhist in general this fundamental good is clearly the cessation of suffering. For Vasubandhu, as for other Buddhists, this is identical with awakening, which itself consists in knowing reality as it is. That is to say, if we are prepared to evaluate the relative worth of our purposes themselves, and re-conceive how we individuate accordingly, those teleologically constructed conceptual systems are to be preferred which more promote the attainment of enlightenment, or the knowledge of ultimate reality.

It is not a matter of indifference whether our ways of conceptualizing and organizing experience and our interactions create more suffering or less, whether they inhibit or promote the ability for all beings to know ultimate reality as it is and thus to eliminate suffering. While the final and more famous Buddhist goal may be the deconstruction of all categories of being and all conceptualizing, it is well recognized and frequently reiterated that astute use of the conventional is indispensable to reaching the ultimate – how we think of everyday life matters because only through this can we hope to transform experience into something less pervaded by suffering.


These are properly ethical advantages to Vasubandhu’s atomism which elude Democritean proto-skepticism.Footnote57 As with Democritus, nothing of how we ordinarily think of and describe the world is correct. None of our words or concepts capture how things really are, nor could they. But as we did not start needing to explain the possibility of change in the face of Parmenidean monism, we do not find ourselves having to defend the validity of our senses, or the possibility of reasoning without them. At the same time, the view does not bleed into an accidental skepticism. Real reality is still (i) knowable, and (ii) approachable. And this matters ethically, not just for the ultimate end of transformative wisdom, but also for more immediate need to determine what we should do with the complex experiences we do have, how we might relate to them, evaluate and change them.

Vasubandhu’s atomism is arrived at through strictly metaphysical and logical considerations about unity, multiplicity, and identity. This gives rise to a distinction between ultimate and conventional reality on which the conventional is not simply hopelessly misleading; it can itself be understood and be used in order to attain understanding of ultimate reality. So far from leading to an inadvertent scepticism, the conception of ordinary reality as constructed from ultimately real elements according to what serves our (equally constructed) purposes indicates knowledge practices which can give us practical and specific information needed for improving that experience. And the prospect of a fully knowable ultimate reality gives guidance on which alterations might count as genuine improvements.

Whether knowledge of ultimate reality itself is a good beyond this is a separate question; the prominence given to prajñā in many Buddhists texts suggests that it is. What this current study shows is how the good of knowing reality could be considered equivalent to the acknowledged Buddhist goal of ending suffering. For it is these sorts of reasons for understanding reality in just this sort of way which can claim to play a role in improving daily experience, and indeed utterly transforming it.


adv. Math. =

Against the Professors


Vasubandhu, Abhidharmakośabhāṣya [Treasury of Abhidharma, with Commentary]


Diels, rev. Kranz, Die Fragmente der Vorsokratiker


Kirk, Raven, and Schofield, The Presocratic Philosophers


Majjhima Nikāya [Middle-Length Discourses]


Milindapañha [Questions of King Milinda]


Pali Text Society


Saṃyutta Nikāya [Connected Discourses]


I would like to thank Ugo Zilioni, whose invitation to participate in a conference on atomism first prompted this work; and also the conference participants themselves, particularly David Sedley, whose contributions offered a valuable perspective on Democritus and Vasubandhu. My thanks are also due to Oren Hanner, whose invitation to participate in a conference on skepticism provided the opportunity to investigate the ethical dimensions of atomism which this paper addresses; and again the conference participants themselves, particularly Mark Siderits, were invaluable in sharpening my arguments. Audiences at the Universität Paderborn, Uppsala Universitet, Boston University, and Columbia University were terrific interlocutors, whose questions have helped to focus and clarify the ideas presented here, and Sylvia Berryman and Ugo Zilioni offered helpful comments on the penultimate draft. Nicholas Lua provided invaluable research assistance.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information


This work was supported by provided the Ministry of Education, Singapore, material support, through research grant number R-607-263-215-121; and by the Templeton Religion Trust, with a fellowship under the auspices of the Beacon Project.


1 Jonathan Barnes, The Presocratic Philosophers, 342. The following quotations are from pages 345, 345, and 349 of the same.

2 And this is so even if one does not share Barnes’ own dismissive view of the very possibility of a meaningful connection between Democritus’ ethics with his metaphysics (see Presocratic Philosophers, 533–4).

3 Aristotle in de caelo Γ4, 303a5 reports that Democritus and Leucippus “say that their primary magnitudes are infinite in number and indivisible in magnitude” (KRS 577).

4 Whether atoms have weight is contested. Aristotle attributes weight to the atoms at de gen et corr. A8, 326a9; and Barnes claims “ample evidence” speaks in favour (DK 68A60, 61, 135), though “orthodoxy now lies with Aëtius”, against. For discussion see KRS ad 573–6, pp. 421–3.

5 “There is an infinite number, and they are invisible because of the smallness of the particles” writes Aristotle, de gen. et corr. A8, 325a30–31 (KRS 545) – though reports differ (for discussion see KRS, pp. 415–6).

6 In Metaphysics Α4, 985b14–15, Aristotle says that the “differences [between the atoms] are three – shape, arrangement and position” (KRS 555), though shape may be expected to take size within its compass; Simplicius, for instance, reports that the atoms “have all sorts of forms and shapes and differences in size” (de caelo 295, KRS 556, DK 68A37).

7 Aristotle, Metaphysics A4, 985b7 “full and solid”; “indivisible and impassive”, according to Simplicius de caelo 242.18 (KRS 557, DK 67A14)

8 Simplicius de caelo 242.21 (KRS 584, DK 67A14); or as Aristotle has it, atoms colliding and associating “are the causes of other things” (Metaphy. Α4, 985b13).

9 See also Aristotle's account in de gen. et corr. A8, “from what is truly one no plurality could come into being, nor a unity from what is truly a plurality - that is impossible” (KRS 545); and in de caelo “the many does not come from one nor one from many” (KRS 479).

10 On the authority of Aristotle, de gen. et corr. A8, 325a2. Kirk, Raven, and Schofield comment that “Leucippus was generally agreed to have evolved his theory of atoms in answer to the Eleatic elenchus” (KRS, p. 403); “Leucippus or Melitus had associated with Parmenides in philosophy” (KRS 539). This is not to say that Protagorean subjectivism had no influence on Democritus’ development of his position (see below), as discussed for instance by Mi-Kyoung Lee in Epistemology after Protagoras. On Eleatic monism see note 12, below.

11 KRS comment (p. 408), “The atomists rejected Zeno’s attempt to show that the members of a plurality are infinitely divisible, and therefore subject to absurd consequences”.

12 While Simplicius credits Parmenides with the view that reality is monoeides and indivisible (Simpl., in Phys. 145.1–146.25, DK28B8), strong monism may in fact be more Melissan than Parmenidean (so Barnes, 204–7), and Zeno is especially associated with paradoxes arising from plurality. However, Plato strongly associates this notion with Parmenides in his dialogue by that name, and the tradition since Aristotle associated Parmenides with the rejection of plurality on the basis that reality is one (Aristotle, Metaphysics 986b29).

13 KRS comment (p. 408), “It is curiously hard to find a text which explicitly calls the atoms uncreated and imperishable, although this is implied by the frequent description of atoms and void as elements and principles, e.g. 555”.

14 νόμῳ γάρ φησι γλυκὺ καὶ νόμῳ πικρόν, νόμῳ θϵρμόν, νόμῳ ψυχρόν, νόμῳ χροιή´ ἐτϵῇ δὲ ἄτομα καὶ κϵνόν. See also DK 68A49 and DK68B125. Barnes prefers Plutarch’s version in adv. Col.1110Ε, which adds the generality “and every combination (sunkrisin)”. This may make the difference as to whether we read Democritus as a reductionist, or as an eliminativist (as argued by Eleni Kechagia in Plutarch Against Colates, Chapter 6).

15 Sedley’s argument on this point, in “Why Aren’t Atoms Coloured?”, is persuasive.

16 Sedley (“Why Aren’t Atoms Coloured?”, 68–69) finds such evidence in Galen, although he judges the tradition’s association of nomisti with the verbal root to have been mistaken.

17 Adv. Math. VII.139; Sextus carries on, “Then, by way of judging [προκρίνων] the genuine one superior to [ἐπιφέρϵι] the bastard one, he adds these words: ‘when the bastard one is no longer able either to see in the direction of greater smallness, nor to hear or smell or taste or sense by touch other things in the direction of greater fineness’” (translation by Sedley in “The Atomist Criteria of Truth”).

18 Judging in Republic V is more obscure (σκοτωδέστϵρον, 478c, 479c) compared to the clarity of knowing, and in Republic VI pertains to sensibles (ὁρατόν, 509d4; see also 510b4–5, 510d6–511a2, 511a8–10) as opposed to the intelligible (νοητόν, 509d4; τὸ γνωστόν, 510a9).

19 Notice how Republic V, a locus classicus for Plato’s distinction between superior and inferior cognition, describes inferior cognitions as being of “the many nomina of the many”, (τὰ τῶν πολλῶν πολλὰ νόμιμα, Rep. 479d2–3). Burnyeat writes, “there is plenty of evidence that Democritean Atomism was based on a priori reasoning, not on observation” citing “clear evidence that his epistemology had a thoroughly rationalist character” (“‘All the World’s a Stage-Painting’”, 66).

20 DKB207, tr. James Warren, with apt discussion, in Democritean Ethics, 48–51. DK68B74 similarly distinguishes pleasantness from benefit.

21 Consider further DK 68B264 for Socratic sentiments about shame before oneself; and DK 68B252 for Platonic (and very unEpicurean) views about the importance and priority of civic responsibilities. Vlastos recognises these Socratic-Platonic elements of Democritus’ ethics, in “Ethics and Physics in Democritus”; and although he insists that “the contrast [of Democritus] with Socrates and Plato remains unbridgeable” (582), he concludes by observing that “Sextus’ association of the materialist Democritus with the idealist, Plato, in opposition to Protagorean phenomenoalism is profoundly true” (592). As we shall see below, however, it is not so easy for Democritus to avoid the pull towards Protagoreanism, just inasmuch as the contrast with Platonic metaphysics and epistemology remains unbridgeable.

22 James Warren’s Epicurus and Democritean Ethics (Chapter 2, passim, esp. p. 72) is especially wise in its discussion of how many contrary positions might be legitimately supported by plausible interpretations of Democritus’ ethical remarks. Retrospectively pinning a specific ethical view definitively onto Democritus is made still more difficult by the uncertainty over exactly which of the surviving ‘Democritean’ texts are indeed by Democritus.

23 Plato acknowledges a version of this worry in the Parmenides’ knowledge paradox, the separation argument which purports to show that sensibles and intelligibles can have no bearing on one another (Parm. 13a–135c), discussed in my “Separation Anxieties”; see also Sandra Peterson, “The Greatest Difficulty”. As Burnyeat observes, “Democritus does not think that appearances give us a sight or grasp (katalēpsis) of things unseen … the senses do not lead you on, there is no such thing as seeing the four columns as implying more of the same sort” (“‘All the World’s a Stage-Painting’”, 67).

24 David Sedley has argued that Democritus’ presumed authority and distrust of the senses need not be contradictory – but it would leave him a skeptic of an empiricist sort. As Sedley puts it, “Democritus could quite consistently hold both that the senses do indeed command the evidence available to the mind, and that we know nothing for certain, because the senses are themselves unreliable” (“Atomist Criterion of Truth”, 38). Sedley credits Myles Burnyeat with the observation that the two assertions can be consistent.

25 “We neither perceive ‘real reality’ (atoms and void), nor even macroscopic objects and their properties (for example, a square tower),” writes Katja Vogt (“Ancient Skepticism”). “Democritus seems to have argued along these lines (SE M 7.135–9; cf. fr. 9, SE M 7.136; Theophrastus, De Sensibus 2.60–1, 63–4), and accordingly his atomist view of perception can be seen as grounding a kind of proto-skepticism”.

26 Democritus, Sextus relates, writes “in the text On the Forms, ‘With the help of these rules, man should realise [γιγνώσκϵιν] he is far from truth’. And again, ‘This discussion too shows that we in reality [ἐτϵῇ] know [ἴσμϵν] nothing about anything, rather for each there is a reconfiguring – a belief [ἐπιρυσμίν]’. And further, ‘Indeed it will be clear that it is not possible to know [γιγνώσκϵιν] what each thing is in reality [ἐτϵῇ]’. Here he puts nearly every possibility of knowledge in question, although he primarily refers only to sensory perception” (adv. Math VII.137, my translation).

27 According to some of his successors, even Plato did not avoid it, animated as he was by similar distinctions and concerns.

28 Indeed, in texts that go through and beyond atomism – e.g. the 10th C. syncretist Śāntarakṣita relies much on the principle that something cannot be both one and many.

29 See Willemen in Sarvāstivāda Buddhist Scholasticism for the association of Vasubandhu with the Mūlasarvāstivādins (who Willamen identifies, controversially, with the Sautrāntikas), as opposed to the rival Sarvāstivādins, based nearby.

30 Translated by Bhikkhu Bodhi in The Connected Discourses of the Buddha. All quotations of the Connected Discourses are drawn from this source.

31 Probably, although there is only the one text that actually adds ‘and so with every mixture’ (see note 14), rather than focusing specifically on sensible properties as the ‘merely considered so’.

32 Democritus seems to have thought there was a soul, but it is not clear he had a good atomist account of this. Was it an agglomeration of atoms, or a single atom? The evidence of Aristotle tells in favour of a heap of spherical fiery atoms (de Anima I.2, 403b30–404a9), but the implication that a soul is therefore real only by opinion seems not to have been drawn. Nor did Democritus seem to confront the difficulty of a multiplicity of soul atoms accounting for the necessary unity of the mental in cognition (this was a point on which the non-Buddhist Nyāya philosophers pushed the Buddhist no-self theory particularly trenchantly). In his accounts of cognition and his ethics, Democritus seems rather to have helped himself to a unity of soul which his metaphysics ought to have undermined.

33 And although Diṅnāga will later define perception as that which is free from conceptual construction (Pramāṇasamuccaya I.i.3c and I.i.6ab) – and although perception is the superior pramāṇa throughout Buddhist thought, while inference is tainted by conceptualizing and is, therefore, distorting – still, this epistemological allegiance to perception was not the reason for positing dharmas as simples and ultimately real in the first place. Simple ultimately real constituents of reality seem to have been driven instead by logical (what is different cannot be the same) and metaphysical considerations (as the mereological reductionism of Milinda’s chariot, below).

34 See Siderits, Buddhism as Philosophy, 53–6, for analysis of this passage in these terms; and Carpenter, Indian Buddhist Philosophy, 35–47.

35 Carpenter, Indian Buddhist Philosophy, 40–2, examines the apparent slippage between ‘not identical to all of its parts’ and ‘not identical to some subset of its parts’, and offers an argument for why the latter, apparently more plausible option is also unsatisfactory.

36 The distinction between the ‘two truths’, as they are called, fundamentally frames Sanskrit Buddhist philosophy and its successors, with different philosophers drawing the distinction in different ways. Sonam Thakchoe, “The Theory of Two Truths in India”, and Guy Newland, Appearance and Reality, both offer overviews and exposition of this contrasting pair, informed significantly by Tibetan doxographers.

37 Trenton Merricks observes (in conversation, UVa, 20 Nov. 2020) that it is not this principle alone which does the work, but this principle plus a rejection of a building principle (as Karen Bennett calls them, in Making Things Up). Since any building principle (e.g. the constitutes relation, the composes relation, the inherence relation) is tantamount to an assertion of a multiplicity that it is indeed a real unity, I do not think the rejection of composition is anything over and above insisting that one cannot be many, thus putting the onus on any defender of a purported principle of composition to explain how it could be otherwise. In this debate, there are no direct arguments for or against the validity of any such principle: The Abhidharma Buddhist, like Theodor Sider (“Against Parthood”) will appeal to parsimony; their opponent to explanatory power. (In the contemporary discourse, the anti-nihilist may also point to the nihilist’s reliance on the appeal to ‘constituents arranged chair-wise’; but the Ābhidharmika is on firmer ground here, since they do not admit that the chariot-wise arrangement of simples is itself ultimately real). However, considerations of why some subset of chariot parts – let us say, those essential to its definitive function (an essentially Aristotelian option) – are not the real essence of the chariot are found in Carpenter, Indian Buddhist Philosophy, 40–3. My “Persons Keeping Their Karma Together” considers a minority Buddhist position which did seem to think that organismal unity calls for some additional explanatory principle; Vasubandhu argues against this Buddhist Personalist position in his “Treatise on the Negation of the Person”, traditionally found as Abhidharmakośabhāṣya IX, and available in a useful contemporary translation by Kapstein as Chapter 14, Part I, of his Reason’s Traces.

38 I call this a ‘correlate’ because there are good reasons to be cautious about simply identifying them – not least because the appearing quality is not an apt way of distinguishing one side of the Buddhist distinction from the other.

39 This and all translations of Vasubandhu’s Abhidharmakośabhāṣya are adapted from Pruden’s translation, with modifications by reference to Pradhan’s Sanskrit edition.

40 See Cox, “From Category to Ontology” for discussion of the difference between these (and their contrast terms paramārthasat and dravyasat) and the evolution in early Buddhist philosophy from the one to the other, and towards conflating them. Karunadasa’s “The Dhamma Theory” describes how the canonical Abhidharma text, the Dhammasangani, already elaborates samutti as conceptual (paññatti/prajñapti), and how even in early Buddhism the distinction between ultimate and conventional “distinguishes between those types of entities that truly exist independently of the cognitive act and those that owe their being to the act of cognition itself” (20).

41 I analyse Vasubandhu’s atomism in detail, including his rejoinder to the Problem of Contact, in “Atoms and Orientation”.

42 Goodman, “The Treasury of Metaphysics”, offers detailed philosophical examination along these lines, though his further claim that Vasubandhu’s is a ‘two-tiered’ ontology is neither textually nor argumentatively warranted (see “Atoms and Orientation”, notes 16 and 18 for details). For Vasubandhu as offering a trope theory, see also Siderits, “Buddhist Reductionism”; and Ganeri, Philosophy in Classical India, 101–2.

43 In brief: simples cannot decay (since that would imply parts), but only either exist or fail to exist. A non-existence cannot be created through external agency, so the cause of going out of existence must belong to the simple itself. But a simple cannot gradually ‘actualise’ different parts of itself or powers any more than it can gradually decay. Therefore this internal power to cause its own destruction must be fully realised upon the moment of the atom’s arising. Therefore, any simple must have strictly momentary existence.

44 It also underscores the rejection of substance-property metaphysics implicit in dharma-theory (on which, see Williams, “On the Abhidharma Ontology”), as well as the way in which dharmas are more event-like than substance-like (on which, see Warder, “Dharmas and Data”, especially pp. 275 and 290).

45 The precise argument here is obscure, and the dialectical relationship between the Twenty Verses and the Abhidharmakośabhāṣya is complicated. But here we may take the Twenty Verses passage to be putting forward for its own purposes the view that Vasubandhu has articulated (apparently) in his own voice in the Abhidharmakośabhāṣya.

46 Nāgārjuna famously challenged whether this distinction could coherently be made, arguing that individuation itself is always dependent on contrasted ‘others’, and due to mental activity (see Carpenter, “Dependent Arising”). Vasubandhu does not seem to feel the challenge is a serious one.

47 In fact, Democritus’ distinction between sensation and intellection as sources respectively of mis-information and information takes as granted and unproblematic the existence of the body, and this body distinct from others – e.g. “In the Confirmations, although he had promised to assign the power of assurance to the senses, he is nonetheless found condemning them, for he says, ‘But we in actuality grasp nothing precisely as it is, but rather as it shifts according to the condition of the body and things entering and pressing upon it’” (Sextus Empiricus, adv. Math. VII.136 (=DK 68B9) modified translation of KRS 553).

48 Unless, of course, it leads to Platonism – which is just the assertion that the mind does indeed give us access to non-sensory reality as it is.

49 Strictly speaking, so could Democritus’ atomism, if it were taken in a non-skeptical Platonic direction, with a robust account of the intelligible and intelligibility – and essentially left off being atomist. (That is to say, Democritus must give his non-bastard mode of cognition some appropriate objects to cognise). This is certainly not a lineage that either the subsequent atomist or skeptical traditions, or Plato himself, recognized.

50 For instance, the sort of cultural relativism we might worry McDowell’s or MacIntyre’s views lead to by tying the very meaning and intelligibility of concepts to our shared practices. For McDowell, see his “Virtue and Reason” and “Two Sorts of Naturalism”; for MacIntyre see his After Virtue, especially Chapters 14 and 15. The spectre of the dismal slough of relativism is a crucial area of intra-Buddhist debate, as certain Madhyamaka Buddhist views seem unable to retain such a prospect of knowing a distinct ultimate reality and thus are in danger of pernicious relativism; on this, see Tillemans, “How Far Can a Mādhyamika Buddhist Reform Conventional Truth?”. For the record, it may be that the historical Protagoras’ relativism was in fact of the sophisticated sort, rather than the capricious individual sort that Plato first characterizes it as in the Theaetetus.

51 See the prefatory verse and first two verses of Vasubandhu’s Abhidharmakośabhāṣya for his articulation of the claim, which I discuss in “Explanation or Insight?”.

52 I argue for the importance of such an impersonal, unworldly ideal in “Ideals and Ethical Formation”, and explore the associated psychological implications in “Explanation or Insight?”.

53 Witness the nature of many Buddhist meditational exercises, especially various analytic practices. Such meditational exercises were considered indispensable mental cultivation, and essentially salutary.

54 For instance, Buddhaghosa in Visudhimagga IX, and Śāntideva in Bodhicaryāvatāra VI (for discussion of which, see, Carpenter, “Ethics Without Justice”).

55 Is ‘minimisation of crime’ the goal? If so, does conceiving of individuals as divorced from their social context, and building practices of accountability on ascriptions of an internal autonomous will, actually reduce crime?

56 Note that since this question only arises upon properly grasping the impersonal, non-substantial and processive nature of reality, it only arises as a question when the interpretation of it as ‘what is good for me?’ no longer makes sense.

57 These ethical advantages even survive what one might think of as the ‘creeping skepticism’ of idealism. This would be a longer tale to tell. But Vasubandhu himself pushes Abhidharma Buddhism towards idealism; and yet in his idealist text, the Twenty Verses, with Commentary, he offers glimpses at Verses 8–10 of how the transition to full-blown idealism retains the ethical practices and advantages of discerning and analysing conventional reality. Moreover, Vasubandhu’s Yogācāra, both here and as articulated for instance in the Thirty Verses, retains a firm distinction between (realisation of) ultimate reality and conventional cognition. Indeed retaining an ultimate reality that was not conventional was something for which the Mādhyamika Candrakīrti could not forgive Yogācāra Buddhism.


  • Barnes, Jonathan. The Presocratic Philosophers. London: Routledge & Kegan Paul, 1982.
  • Bennett, Karen. Making Things Up. New York: Oxford University Press, 2017.
  • Burnyeat, Myles. “‘All the World’s a Stage-Painting’: Scenery, Optics, and Greek Epistemology”. Oxford Studies in Ancient Philosophy 52 (2017): 33–75.
  • Carpenter, Amber D. “Persons Keeping Their Karma Together”. In The Moon Points Back: Analytic Philosophy and Asian Thought, edited by J. Garfield, G. Priest, and K. Tanaka. New York: Oxford University Press, 2015.
  • Carpenter, Amber D. “Ethics Without Justice: Eliminating the Roots of Resentment”. In A Mirror is for Reflection: Understanding Buddhist Ethics, edited by Jake Davis. New York: Oxford University Press, 2017.
  • Carpenter, Amber D. with Sherice Ngaserin Ng, “Atoms and Orientation”. In Atomism in Philosophy, edited by Ugo Zilioli, 159–181, Bloomsbury Academic, 2020.
  • Carpenter, Amber D. “Nāgārjuna: Dependent Arising Without Any Thing Arising”. In The Routledge Handbook of Indian Buddhist Philosophy, edited by William Edelglass, Sara McClintock, and Pierre-Julien Harter, 211–235. Routledge, 2023.
  • Carpenter, Amber D. “Separation Anxieties – Parmenides 133a–35c”. In Plato’s Parmenides, edited by L. Brisson, A. Macé and O. Renault, 245–252. Brill, 2022.
  • Carpenter, Amber D. “Ideals and Ethical Formation: Confessions of a Buddhist-Platonist”. In Reasons and Empty Persons: Mind, Metaphysics, and Morality: Essays in Honour of Mark Siderits, edited by Christian Coseru, 387–415. Springer, 2023.
  • Carpenter, Amber D. “Explanation or Insight? Competing Transformative Epistemological Ideals”. In Crossing the Stream, Leaving the Cave, edited by Carpenter and Harter, Oxford University Press, forthcoming.
  • Cox, Collett. “From Category to Ontology: The Changing Role of Dharma in Sarvāstivāda Abhidharma”. Journal of Indian Philosophy 32, no. 5/6 (2004): 543–97.
  • Diels, Hermann, ed., rev. Walther Kranz. Die Fragmente der Vorsokratiker: Griechisch und deutsch. Original edn. 1903. Revised, 6th edn. 1951. Reprint Berlin: Weidmann, 2005.
  • Diṅnāga. Pramāṇasamuccaya I, translated with commentary by Masaaki Hattori, as On Perception: Being the Pratyaksapariccheda of Dignaga’s Pramanasamuccaya.: Cambridge, MA: Harvard University Press, 1968.
  • Ganeri, Jonardon. Philosophy in Classical India: The Proper Work of Reason. London: Routledge, 2001.
  • Goodman, Charles. “The Treasury of Metaphysics and the Physical World”. The Philosophical Quarterly 54, no. 216 (2004): 389–401.
  • Horner, I. B., trans. Milinda’s Questions. 2 vols. London: Pali Text Society, 1963–4.
  • Karunadasa, Y. “The Dhamma Theory: Philosophical Cornerstone of the Abhidhamma”. The Wheel Publication 412, no. 413 (1996): 1–67.
  • Kapstein, Matthew. Reason’s Traces. Boston: Wisdom Publications, 2001.
  • Kechagia, Eleni. Plutarch Against Colates: A Lesson in History of Philosophy. Oxford: Oxford University Press, 2011.
  • Kirk, G. S., J. E. Raven, and M. Schofield. The Presocratic Philosophers: A Critical History with a Selection of Texts. 2nd ed. Cambridge, UK: Cambridge University Press, 1983.
  • Lee, Mi-Kyoung. Epistemology After Protagoras: Responses to Relativism in Plato, Aristotle, and Democritus. Oxford, UK: Oxford University Press, 2005.
  • MacIntyre, Alasdair. After Virtue: A Study in Moral Theory. 3rd ed. Notre Dame, IN: University of Notre Dame Press, 2007.
  • Majjhima Nikāya. [The Middle Length Discourses of the Buddha: A Translation of the Majjhima Nikāya]. Translated by Bhikkhu Ñāṇamoli and Bhikkhu Bodhi. 4th ed. Somerville, MA: Wisdom Publications, 2009.
  • Mansfeld, Jaap. Die Vorsokratiker: Auswahl der Fragmente, Übersetzung und Erläuterungen. Philipp Reclam: Stuttgart, 1987.
  • McDowell, John. “Virtue and Reason”. The Monist 62, no. 3 (1979): 331–50.
  • McDowell, John. “Two Sorts of Naturalism”. In Virtues and Reasons: Philippa Foot and Moral Theory; Essays in Honour of Philippa Foot, edited by Rosalind Hursthouse, Gavin Lawrence, and Warren Quinn, 149–80. Oxford, UK: Clarendon Press, 1998.
  • Newland, Guy. Appearance and Reality: The Two Truths in Four Buddhist Systems. Ithaca, NY: Snow Lion Publications, 1999.
  • Peterson, Sandra. “The Greatest Difficulty for Plato’s Theory of Forms: The Unknowability Argument of Parmenides 133c–134c”. Archiv für Geschichte der Philosophie 63 (1981): 1–16.
  • Pradhan, P. Abhidharmakośabhāṣyam of Vasubandhu. Patna: K. P. Jayaswal Research Institute, 1975.
  • Pruden, Leo. Abhidharmakośabhāṣyam of Vasubandhu. A translation into English from the French translation of Louis de la Vallée Poussin. Berkeley, CA: Asian Humanities Press, 1988–1990.
  • Saṃyutta Nikāya. [The Connected Discourses of the Buddha: A Translation of the Saṃyutta Nikāya]. Translated by Bhikkhu Bodhi. Somerville, MA: Wisdom Publications, 2000.
  • Sedley, David. “Sextus Empiricus and the Atomist Criteria of Truth”. Elenchos 13 (1992): 19–56.
  • Sedley, David. “Why Aren’t Atoms Coloured?”. In Atomism in Philosophy: A History from Antiquity to Present, edited by Ugo Zilioli, 61–75. London: Bloomsbury Academic, 2020.
  • Sider, Theodore. “Against Parthood”. Oxford Studies in Metaphysics 8 (2013): 237–93.
  • Siderits, Mark. “Buddhist Reductionism”. Philosophy East and West 47, no. 4 (1997): 455–78.
  • Siderits, Mark. Buddhism as Philosophy: An Introduction. Aldershot, England: Ashgate, 2007.
  • Thakchoe, Sonam. “The Theory of Two Truths in India”. In Stanford Encyclopedia of Philosophy. Stanford University. Originally published February 17, 2011; substantively revised October 20, 2022. https://plato.stanford.edu/entries/twotruths-india.
  • Tillemans, Tom J. “How Far Can a Mādhyamika Buddhist Reform Conventional Truth?: Dismal Relativism, Fictionalism, Easy-Easy Truth, and the Alternatives”. In Moonshadows: Conventional Truth in Buddhist Philosophy, edited by The Cowherds, 151–65. Oxford: Oxford University Press, 2011.
  • Vasubandhu. Abhidharmakośabhāṣya, edited by P. Pradhan. Patna: K. P. Jayaswal Research Institute, 1975.
  • Vasubandhu. Abhidharmakośabhāṣya, translated into English by Leo Pruden from the translation into French by Louis de La Vallée Poussin. 4 vols. Berkeley, CA: Asian Humanities Press, 1988–90.
  • Vasubandhu. Twenty Verses, with Commentary. Edition and Translation by Stephen Anacker in Seven Texts by Vasubandhu, The Buddhist Psychological Doctor. Delhi, Motilal Banarsidass 1984, rev. 2005.
  • Vlastos, Gregory. “Ethics and Physics in Democritus”. Philosophical Review 54, no. 6 (1945): 578–92.
  • Vogt, Katja. “Ancient Skepticism”. In Stanford Encyclopedia of Philosophy. Stanford University. Originally published 2010; substantively revised 2018. https://plato.stanford.edu/entries/skepticism-ancient.
  • Warder, A. K. “Dharmas and Data”. Journal of Indian Philosophy 1 (1971): 272–95.
  • Warren, James. Epicurus and Democritean Ethics: An Archaeology of Ataraxia. Cambridge, UK: Cambridge University Press, 2002.
  • Willemen, Charles, Bart Dessein, and Collett Cox. Sarvāstivāda Buddhist Scholasticism. Leiden: Brill, 1998.
  • Williams, Paul. “On the Abhidharma Ontology”. Journal of Indian Philosophy 9 (1981): 227–57.

Reprints and Corporate Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

To request a reprint or corporate permissions for this article, please click on the relevant link below:

Academic Permissions

Please note: Selecting permissions does not provide access to the full text of the article, please see our help page How do I view content?

Obtain permissions instantly via Rightslink by clicking on the button below:

If you are unable to obtain permissions via Rightslink, please complete and submit this Permissions form. For more information, please visit our Permissions help page.