From complexity to emergence and beyond ...

For a different example of issues in emergent systems and their control, see Józef Okulewicz's The Structural Formulation of Logistics.

Several interesting articles recently ...

On the "reality" of indeterminacy ...

Losick, R. and Desplan, C. (2008) Stochasticity and cell fate. Science 320: 65-68.

Faisal, A. A., Selen, L.P.J., and Wolpert, D.M. (2008) Noise in the nervous system. Nature Reviews Neuroscience 9: 292-303.

See also Grobstein, Paul (1994) Variability in behavior and the nervous system. IN: Encyclopedia of Human Behavior, Vol. 4 (V.S. Ramachandran, ed). NY: Academic Press. pp 447-458.

And on social organization ...

How to turn a herd on Wall Street, NY Times Week in Review 6 April 2008

Interesting conversation in the emergence group this morning, intersecting in intriguing ways with the philosophy of science course I'm currently co-teaching.

A persistent problem in understanding emergence is the lack of models that compellingly illustrate the development of several distinct levels of organization, a phenomenon that seems quite compelling in describing the world (eg cells, multicellular organisms, societies of multicellular organisms). While one can readily demonstrate "emergent" properties at one level resulting from interactions of elements at a lower level that don't display those properties, it seems difficult to "propagate" emergence. This may relate to the fact that detection of "emergent" properties in models so far depends on an outside observer who makes the distinction between lower level and higher level properties. The two sets of properties aren't clearly distinguishable within the systems (models) themselves.

This problem is, I think, closely related to the distinction between "weak" and "strong" emergence, ie to the question of whether higher order properties are or are not "fully explained" by the properties of lower lower elements and their interactions. In the past, I've argued that that there are in fact strong emergence situations, ones is which higher order properties are not "fully explained", because lower order properties and interactions "permit but do not account for" higher order properties, that there is some essential "information addition" that occurs in going from one to the others. A particular instance of this would be a situation where either lower order properties or interactions have some degree of indeterminacy involved with them.

I continue to think that "information addition" and, in particular, "indeterminacy" is an important aspect of the full power of emergence. But it hasn't been and continues not to help me think about the levels of organization problem. What may do so is the addition of "top-down causation". In effective descriptions of the world, higher order phenomena not only emerge from interacting elements not displaying those phenomena but also become causally significant, ie they influence lower order elements, causing them to behave in ways that they wouldn't in the absence of the higher level organization. This creates a causal feedback loop, one which couples lower and order higher phenomena in ways that make them reciprocally influential, simultaneously both cause and effect. My intuition (incentive for further exploration in models) is that it is precisely this coupling that creates distinguishable levels of organization (ie it is top-down causation and associated coupling that has been absent from most emergence models to date). The existence of "coupling" would provide a way to distinguish higher order properties from lower order ones within systems themselves, without the need to appeal to an observer to recognize the higher order properties.

Whether this turns out to provide a solution to the "propagation" problem or not, it does suggest that "strong emergence" is so not only because of indeterminacy but also because of top-down causal influences. The properties and interactions of lower order elements in systems with vertical loopiness are likely not to be those that would exist in the absence of vertical loopiness. What this means is that the properties and interactions one describes using experimental techniques that isolate the lower level constituents may not be the ones that are relevant for understanding the higher order phenomena one is interested in. The higher order properties are genuinely not reducible to lower order ones because they depend on the existence of new top-down influences that come into being as a part of the emergent process. To put it differently, once a higher level system comes into existence that influences lower level attributes and properties, one can't expect to find by reductionist approaches the lower level properties that are relevant to the overall organization. The lower level and upper level phenomena need instead need to be thought of as "co-evolving".

A good example of this, I suspect, is the relation between the human brain (a lower level phenomenon for present purposes) and culture (a higher level phenomenon). In many social science realms, a reductionist approach is adopted (consciously or unconsciously): to understand collective behavior one characterizes properties of individuals and their interactions (or asserts them, eg homo economicus) and then tries to use those to derive (or "explain") social phenomena. More generally, one asserts the existence of some fixed "human nature", and then uses this as a foundation to make sense of interpersonal dynamics and social organization. The approach can of course be a productive one, but the recognition of top-down causation is also a useful reminder that it may not always work. There probably is, at this point in time, no such thing as "human nature" independently of the social context within which one finds particular individuals (cf also East and West Part Ways in Test of Facial Expression). The brain is a nexus point of "bottom up" and "top down" influences, and one will inevitably find oneself in trouble if one neglects either in trying to make sense either of the brain or of social organization (interacting assemblies of brains).

Something entirely analogous almost certainly goes on within the brain itself. The properties of assemblies of neurons derive in part from the properties of neurons themselves but also in part from the place of those particular assemblies in the larger assembly which is the brain, and so one can be misled about the properties of particular assemblies if one studies them only in isolation. On a slightly larger scale, the neocortex (or "story teller") probably functions as a distinctive source of "top down" influences that derive from but are not totally determined by the bottom up signals it gets from the rest of the nervous system.

Ambiguous figures provide a particularly dramatic example of interactions of bottom up and top down influences. One can and does attribute a variety of different "meanings" to such figures, the "meaning" being the "story" that is settled on (by the neocortex?) as a way to account for an array of bottom up signals that themselves are "meaningless". The "meaning" in turn influences the rest of the nervous system in coordinated ways that produce behaviors appropriate for that meaning.

And this in turn suggests a potentially useful way to think about "meaning" itself. Things don't have "meaning" in and of themselves but only have it insofar as they are given meaning by an agent having the capacity to do so. "Meaning" is a top down construct, one not fully determined by bottom up influences, and therefore always containing (if one is aware of it) an element of choice. At the same time, "meaning" clearly has causal significance in that it can/does significantly influence other aspects of the nervous system and so behavior. Its an intriguing thought that the existence of "meaning" for other things (and for oneself?) depends fundamentally on the potential to change it, to entertain the possibility of alternative higher order objectives. In any case, clearly "meaning", like molecules, cells, communities, is derivative of other things but also has some independent existence and causal significance.

Meaning having "some independent existence"? Let me be clear about that, lest I be accused of resurrecting a dualist perspective. "Meaning", in the present sense, is not a second realm of things, parallel to a material world. It is no more (and no less) than organized matter, eg a pattern of neuronal activity in a brain. It does though have "some independent existence" for two reasons. One is that it does not follow necessarily from other organized states of matter at lower levels of organization (because of indeterminacy and the loopiness discussed above), and has causal efficacy. In addition, a given "meaning" may have somewhat different incarnations in different brains (or in the same brain at different times). That they are all in some sense the same is a function of similarities they create in other brains (or the same brain at different times), just as Beethoven's third symphony is the same whether incarnated on a record, a CD, or a musical score. There is no second parallel realm of "meaning" (or more generally "ideas") here; there is though a recognition that meaning (and ideas), while they have their origins in organized matter, become causally significant in ways that are not simply and totally accountable for in terms of organized matter. There is no "meaning", no "idea" independent of organized matter, but organized matter can and does bring into existence "meaning" and "idea" as a somewhat indeterminate and causal influence on organized matter, particularly (but not exclusively) on those forms that create and respond to "meaning" and "idea".

All of this seems to me useful in further clarifying emergence, and most particularly in thinking about the kinds of hybrid systems that emergence gives rise to, those in which local distributed processes interact with more global and architect/designer like ones. But it bears in interesting ways on philosophy of science, inquiry in general, and individual human lives as well. The idea that "meaning" exists only when it is bestowed on something by an observer, and that a given thing can have multiple meanings, seriously challenges the idea that science is "about reality", about "things as they are". What an observation "means" is, and should alway be thought of as being subject to revision. On the flip side, though, the attribution of a particular meaning (from within the array of possible meanings) to an observation is the crack through which an observer can not only acquire information about the world but contribute to shaping it. From this perspective, science (and inquiry in general) is as much about creating "reality" as it is about revealing it. Similarly, one should probably think not so much of "finding" meaning in life as of creating it. That meaning is created rather than found is the opening for hybrid entities to play an active and deliberate role in their own lives and in the worlds they find themselves in.