Friday, January 23, 2009

Reduction or emergence?

How bad is reductive thinking, and how useful is the concept of emergence?

Nothing gets new age folks, theists, and other supernaturalists upset like the reductive scientific approach to knowledge. What could be more dry, stultifying and un-holistic than the practice of smashing a mystery to pieces and figuring out how those mundane pieces fit back together? In a similar vein, some scientists are becoming enamored with "emergence" as a way of describing the novel properties of complex systems that may not be "reducible" to more elementary phenomena, or at least whose relationship to them is unknown or obscure. A proponent of this angle is scientist Stu Kauffman, who sits on the fence of using sacred language while promoting what is at heart a Spinozist, Einsteinian, secular view of reality- that if one must have a god, it should be one of reality in its incomprehensible totality, not an infantalizing, tribalizing, patriarchal, and blood-soaked totem.

One confusion in this debate is the question of whether we can possess truth. As Hume and Kant described, we can not know reality in a comprehensive way. We can model aspects of it more or less accurately, and follow pragmatists like William James in making practical benefit and closeness to concrete reality the metric of a model's accuracy. That is the best we can do. We would like to be omniscient and god-like, but we are not, and no amount of imagining will get us there.

How does one use past events to predict future ones? The first step is of course to understand past events. And we can not do that without breaking them down into their component relationships. Whether one hypothesizes supernatural causes or natural ones, every event in the world has its causes, and our analysis of the past consists of enumerating what these are, both in principle to find out whether such causes are possible, and in practice, to find out whether such causes actually connect to their hypothesized effects. Once all the pieces are laid out and understood, we can build the abstract models we use to "understand" the world. Thus we come full circle, synthesizing what has been previously reduced, into mental models of reality that are (at best) powerful, useful and accurate. This as holistic as we can reliably get.

Ideally, these causal relationships are tidy enough to be described mathematically, like Newton's gravitation. But phenomena like biology are not so tidy, and there we have to make do with more or less localized causes like chemical principles, molecular shapes and interactions, and ecological constraints, combined with the relatively general principle of natural selection. Because of our mental incapacity to grasp reality whole, as well as our lack of the knowledge to even begin that task, grasping it in bits and pieces is the only way forward, and has allowed imposing edifices of practical abstract knowledge to be constructed back up out of those pieces. In short, reductive thinking is hugely successful, just like the practice of breaking down tasks to smaller levels has allowed great complexity to flourish in economic affairs, in software programming, in manufacturing, etc.

There is a huge temptation to believe that we take a shortcut, that we can grasp the whole of reality by force of intuition, through our unconscious mental capacities, or in consulation with ancient gurus possessing esoteric wisdom. It is true that in some areas, such as the social sphere, intuitions can be powerful ways of knowing- faster at any rate than explicit rational capacities. But of course there are many other areas, like the natural world at scales other than our common experience, where this is not true at all, since the evolution of our intuitive faculties failed to account for quantum mechanics, the deep histories of geology and biology, celestial mechanics, and much else. And even where it is powerful, intuition is not dispositive. In the end we have to use reason to tell whether intuition has been accurate, as the current Madoff financial scandal makes so painfully clear for his conned customers. The close relative of social intuition is the intuition of spirits and spiritual "realities". While clearly palpable to many people, its effects never have been and never are verified by reason, and appear to be fundamentally in error. Believers thus must take refuge in faith alone, for what that is worth.

Once such refuge is taken, other forms of knowledge become a threat rather than a boon, making believers nervous about those taking the reductive approach to knowing. Faith requires mystery, which requires ignorance. Those who would reveal and resolve mysteries strike at the very heart of faith, for where can god live but in the gaps of knowledge? What room is there for esoteric and transcendent human (or super-human) potentials in a world where humans are material beings down to their very electrons?

Returning to reductionism, the product of reductive and integrative analysis is still an incomplete view of reality, and will always remain so. Thus the attractiveness of other modes of thought that offer greater vistas at various levels of fatuousness, like holism, Gaia theory, Gnosticism, etc. "Emergence" is a relatively sober term used by scientists to describe phenomena that can not be easily constructed (in our minds, at least) from more elemental phenomena, such as the emergence of weather patterns out of the complexity of atmospheric physics, or the emergence of social networks out of the buzz of the internet.

I think the concept of emergence is however simply a label for ignorance. If we can not build a complete model of how a butterfly's wingbeats affect the weather a year hence, it is not because of a supernatural principle that generates emergent properties out of thin air, as it were, but because we are ignorant- both of the full causal chain that these wingbeats put in motion, and of the overall chaos/noise in the system which may allow this effect to either amplify or damp out to nothing. Chalking a phenomenon up to emergence may be a shorthand for cases where complexity totally overwhelms our capacity to analyze what is going on, but it is not in itself explanatory, since causes are still afoot, and we would be better off trying to understand them rather than labelling them with a word.

A case in point is natural selection- Darwin's mechanism of biological evolution. How does that arise from simpler phenomena? Or must we label it emergent, and leave it at the biological level only? Its principle lies in competition, which exists at all sorts of levels, from chemical to astronomical. On the astronomical scale, the gravitation of massive objects leads to competition for mass accumulation, forming the lumpiness that we see as galaxies, stars, black holes, nebulae, etc.

Chemicals compete automatically as well in their race to lowest energy, reacting with the first possible partners that come along, sucking the most reactive species out of a collection, then the next reactive, and so forth. Such competition is part of the basis for pre-biotic evolution, where vast numbers of organic chemicals, combined with abundant inputs of energy, appear to have generated reactive units of increasing sophistication, both in terms of competing for metabolic resources from the medium, and eventually also in reproducing themselves- the ultimate case of emergence, as it were.



Incidentally, emergence also pops up in the philosophy of consciousness, renamed in this case "emergentism", which goes something like: while the world may be causally closed, complex properties like consciousness can not be understood as made up of more elementary ones (or be reduced to them) but exist de novo, as new empirical facts, understandable only at the appropriate level. It is rather telling that the Wiki page on this topic offers a series of chemical examples that are then acknowledged to be entirely reducible after all.

For another example, the Philosophy Bites podcast interviewed a philosopher of mind (Tim Crane) who attempts to argue this position- for a sort of confusion and ignorance on behalf of our inability to reduce consciousness to brain functions, yet at the same time claims to not be a dualist, soul-ist, or Cartesian. The questions are excellent, the answers much less so. There is absolutely no reason to grant this veil of ignorance to the phenomenon of consciousness, (complex as it is), which is being studied assiduously, with excellent prospects for understanding that will blow away the fog of "emergentism" (see a recent post).

To close, let me offer a quote taken from a podcast interview by Ginger Campbell of one of my scientific heros, Georgy Buzsaki, neuroscientist and author of "Rhythms of the brain" (see side-links) about brainwaves, where he expresses his view of reductionism:
"The complex systems (area of engineering) offered a very rich toolkit for neuroscience to think about interactions in the brain in a new way, but it also was very important to realize that many of these ideas are important because you can view things differently, but ideas and principles that have a common thread across different disciplines are substrate-free. But whenever we want to understand the mechanisms we have to translate these interesting principles into mechanisms on a given substrate. So when I learned, and with other people we tried to say ... 'oh, the brain is a complex system which has particular dynamics and is non-linear' ... all of these things, it didn't tell me anything, as it doesn't tell anything to the average reader, because the mechanisms have to be understood and broken down into pieces. And that's where I think the responsibilities of neuroscientists lie- that the hand-waving, interesting explanations have to be translated into neuronal mechanisms. And this is where I think the new field of neurocomputation and experimental neuroscience must work together to see the ideas that spring out from your brain or from your head can be really tied to reality, rather to just express imagination."



Incidental links:
  • Radiolab audio episode on the sometimes bizarre joys of science.
  • Review of the science vs religion debate in TNR, making special mention of prominent scientists who try to have it both ways, as well as the so-called Intelligent Design movement. This is a specially trenchant and thorough article- highly recommended.