Contextually Coherent Convention
Or, how to construct knowledge in the face of Buddhist deconstructionism (Part I)
Something that is not dependently arisen
Such a thing does not exist.
Therefore a non-empty thing
Does not exist.– Nāgārjuna, The Fundamental Verses of the Middle Way¹
Madhyamaka Buddhism presents an ontology in which all phenomena are said to be dependently originated — i.e., they exist in dependence upon other phenomena — and that it is the fact of their dependent origination that makes them empty of intrinsic nature. In this manner, ultimate reality, which is the emptiness of all phenomena, is understood as identical to conventional reality², which constructed through our conceptions, and hence dependently originated.
Some aspects of such a metaphysics appear fairly intuitive for those of us who subscribe to modern scientific worldviews. After all, the laws of physics seem to imply that all phenomena are brought about by causes and conditions, and the accompanying rejection of a fundamentally existing self is easily understood using the reductionist language of cognitive science. Even the primacy of conceptual construction can be understood with reference to the idea that scientific theories are merely models for how the world works that may not fully represent how things actually are.
Other aspects of this doctrine, however, can come off as perplexing, contradictory, and even nihilistic. One apparent paradox is the purported emptiness of emptiness, but what seems even more concerning than this logical conundrum is the immediate implication that dependent origination is empty too, which is just another way of saying that neither causes nor conditions nor conceptual construction have any ultimate validity. This flies in the face of the common sense intuition that must be a world ‘out there’ out of which we construct conventional concepts, or at the very least some real basis for knowledge, whether such a basis is ontological (as in an external, mind-independent universe) or epistemological (e.g., perception, inference, or the scientific method). Even if scientific models are just models, so the intuition goes, surely they are models based on something.
But as Madhyamaka philosophers would have it, if any such foundation exists, it is at best only conventionally valid. Some schools like the Prasangika could be read as going further still, rejecting even conventionally valid epistemic foundations and hence the possibility of constructive philosophy altogether³. Is this the ultimate lesson to be taken from Madhyamaka philosophy? Possibly. But if it is, then it seems that ordinary beings trapped in samsara and conditioned to accept scientific materialism have little reason to take it seriously, much less be inspired by it to achieve enlightenment. Perhaps then, we can instead find a lesson in the flipside of the ultimate, which is just the conventional, and create room for knowledge out of emptiness, which is just dependent origination.
What follows then, is an attempt to find such a lesson — an attempt at constructive philosophy that demonstrates the possibility of constructive philosophy even under a thorough-going anti-realism and anti-foundationalism. Drawing upon previous attempts⁴, I hope to sketch a theory of conventional truth and reality that is both more complete and more consistent. I further hope to show that this theory is not just compatible, but in fact stands in mutually beneficial relation with many important aspects of contemporary scientific understanding.
In doing so, I would also like the dissolve at least part of the worry behind the 14th Dalai Lama’s pronouncement that “If scientific analysis were conclusively to demonstrate certain claims in Buddhism to be false, then we must accept the findings of science and abandon those claims.”⁵ A Buddhist understanding of science is by no means contradictory or impossible. On the contrary, it may be the only way of understanding science that is truly defensible.
The Apparent Utility of Reductionist Contexts
To allow room for the improvement of epistemic practices within a Madhyamaka framework, Mark Siderits argues for a form of epistemological contextualism — the idea that epistemic instruments, and hence the truth claims they deliver, are only valid with respect to a certain context.⁶ In particular, Siderits argues for a contextualism that can accommodate reductionism in order to account for the apparent utility of reductionist science, such as the benefits of medicine developed through an understanding of microbiology.
Thus, contexts are conceived of as existing in a hierarchy, “so that each is seen as an improvement on its predecessor.” Presumably, a context is an improvement because it is a reduction of its predecessor. This lends itself easily to a description in terms of conventional and ultimate truth; the hierarchy is understood as “an unending series of sets of triples: false conventional, true conventional, true ultimate, where the true ultimate of one set is the true conventional of an adjacent set.”
As intuitive as this might sound, it is worth questioning the motivation behind such a contextual hierarchy. What, after all, accounts for the apparent utility of reductionist contexts? One answer is that they seem to be more complete. By breaking down wholes into parts, we can account for a range of phenomena in terms of the parts that cannot be accounted for in terms of the wholes. Hence, disease can be explained with reference to cells and viruses but not with reference to persons, and wave-particle duality can be explained with reference to quantum mechanics but not with reference to workings of macroscopic classical objects. With an understanding of a wider range of phenomena, we are then better able to address our ends and needs, like the reduction of suffering caused by disease.
It is important to note, however, that completeness need not be all that goes into determining the relative truth of an epistemic context. Furthermore, it is not always evident that the contexts we think of as reductions are in fact more complete. For example, critics of cognitive science often question a computational understanding of the mind, pointing out that the computer metaphor is assumed rather than demonstrated.
Without resorting to transcendental arguments, this criticism is made more palpable by the fact that despite our sound understanding of quantum electrodynamics, computational chemists have yet to fully reproduce to knowledge and predictions of traditional chemists, in part due to the computational complexity of the simulations required. If even chemistry cannot be demonstrated through computation of physical laws, what hope do we have for cognitive science? This can be taken in two ways: either we cannot meaningfully claim, due to the lack of demonstration, that physics is complete (a formal consideration), or even if we can, it is a computationally inefficient⁷ means of understanding the world (a pragmatic consideration).
Even if the formal consideration is countered by arguing that lack of demonstration does not mean lack of compatibility (and so in Popperian mode, we accept physics as a hypothesis until shown incompatible with chemical or neuroscientific facts), the pragmatic consideration still has plenty of force. Indeed, it appears to be a primary reason for the diversity of academic disciplines that exist today; wherefore the economists and sociologists, if physicists could efficiently give us all the answers?
It seems then, that the apparent utility of reductionism cannot be taken for granted. Or, if we want to maintain its utility, we have to appreciate how reductionism can be understood in at least two different (but non-exclusive) ways. The first understanding is mereological; it is to understand a whole in terms of its parts, and it is this understanding of reductionism that is normally associated with the natural sciences, as well as the Abhidharma school of Buddhist philosophy from which Siderits draws inspiration.
The second understanding is computational; it is the reduction of complex phenomena into a more elegant, parsimonious, computationally efficient explanation. In performing such a reduction, completeness may be lost, as in the failure of a homo economicus model in explaining a wide range of human behaviour, but what is gained is the ability to efficiently explain or predict some other range of phenomena that other epistemic contexts are unable to account for.
Applying this understanding to the above then, the kind of reductionism that best seems to explain our epistemic progress since the time of the Buddha is computational reductionism. While mereological reductionism certainly plays a role, and is largely concomitant with its computational cousin in the natural sciences, it fails to the account for the progress we have made in the social sciences. In light of all this, conceptualizing epistemic contexts as arranged in a linear hierarchy seems highly limiting. A more productive approach might to imagine them as a tree, or more generally, a directed graph³, where contexts are analogized as nodes in the graph, and directed edges are thought of as reductions or other sorts of dependence relations.
Before investigating what such a directed graph might look like, one last consideration against a linear hierarchy should be brought to fore. A key aspect of the Madhyamaka doctrine of emptiness is the understanding that all constructs are conceptually imputed. Therefore, an epistemic context is not to be understood as non-ultimate simply because there is a context underneath it that serves as a causal reduction, whether mereological or computational. Importantly, they are also understood as non-ultimate because they are conventional — shaped by human interests and practices.
We have seen this aspect surface in the notion that epistemic contexts are constructed models, which furthermore account for epistemic progress by virtue of their utility, but have not analysed these concepts in depth. In order to present a picture that not only better captures Madhyamaka thought but is also more complete, we have to turn to a mechanistic account of reality as merely conventional.
This is part one of a three-part essay, with parts to be released once a week.
¹Garfield, The Fundamental Wisdom of the Middle Way.
²But epistemically different. c.f. Cowherds, Moonpaths, 2010, chap. 2.
³Cowherds, Moonpaths, 2010, chap. 3,6,9.
⁴Ibid.
⁵Dalai Lama XIV Tenzin Gyatso, The Universe in a Single Atom.
⁶Cowherds, Moonpaths, chap. 10. Is Everything Connected to Everything Else?
⁷Perhaps irretrievably so, if certain simulations are NP-hard.
⁸Which might even contain cycles!
