Huh

I just checked out a great site for election data, fivethirtyeight.com, based on how high it’s listed by Google and the number of comments, I suspect I’m one of the last people to hear about it. Oh well. Anyway, I mention this because I’ve been wondering about early and absentee voting. It seems to me, the longer a candidate leads during early voting periods the harder it is for his/her opponent to catch up. For example, if candidate O leads candidate M by 5 points in a given state during the early voting period I’d expect the early results to represent that spread, assuming early voters are representitive of all voters in that state (historically early voters have been both more partisan and older). Still, if a candidate benefits from early voting it means the lagging candidate needs to corral not only a winning number of votes on election day but by a large enough margin to make up for the number of days s/he was in deficit during the early voting period. While it’s true McCain is doing better amongst older voters, Obama seems to be leading in the enthusiasm race, further, my understanding is, Obama’s campaign has done a better job of organizaing “early voters”. In short, I suspect Obama’s support may be understated by the polls since he has been leading in most swing states for all of the early voting period. We’ll see.

In my last post I relied on an observer to determine whether a system is or is not complex. I have run across a potential work-around. It comes from “A Universe of Consciousness” by Edelman and Tononi. The authors, focused on a neurological explanation for consciousness, posit the necessity of viewing the human brain as a complex adaptive system. Using an interesting twist on information theory they propose measuring complexity using the system itself as the observer.

In effect they take a complex system, isolate it then split it in two . First they measure the entropy of each of the two parts,  then they measure how one part of the system affects the rest of the system and vice versa.  As they say, “If the system is isolated then from the point of view of that element, the only information available is given by the difference in the state of the rest of the system that makes a difference to the state of the element.” Their approach uses a statistical methodology to measure to what extent the entropy of the overall system is accounted for by the entropy of the subpart and vice versa. They argue high values of complexity correspond to an optimal synthesis of  functional specialization and functional integration within a system. For example, a crystal (as they point out) is functionally integrated but not complex due to a lack of functional specialization. Meaning all the molecules are tightly connected but there is zero molecular specialization, in effect, nothing happens the system is in a state of equilibrium. While a gas would be low in complexity due to a lack of integration, lots of stuff going on but what one molecule does will have very little affect on any other random molecule. In this case the system is chaotic. Now, I have to admit, I’m bothered by their use of the word “optimal” but, by effectively making the system it’s own observer they seem to have minimized the need for an external observer and potentially offered a quantitative measurement of complexity.

However, even if one accepts the above approach as a measurement of complexity I don’t think it addresses the need for an external observer to measure Emergence. My understanding is Emergence is different than complexity or rather Emergence is born of complexity. A new layer or pattern comes into being due to the complexity (is this the only criteria?) of the underlying layer. But, the ant hill (to take a typical example of emergence) is only seen as an emergent pattern if identified by the external observer. In fact, the ant hill itself, does not seem to meet the definition of complex as given above. Perhaps it could be called a “ceiling pattern” since the ant hill does not seem to give birth to any further complexity and therefore there’s no further evidence of Emergence. As I write this I find myself wondering if one might argue that there are two kinds of emergence, a “ceiling pattern” or perhaps non-complex emergence and Emergent Complexity. Perhaps the difference is, not only that Emergent Complexity gives birth to more layers of complexity but also it has the potential to be its own observer.

I’ve started this post eight different times, each time I tried a different tact and each time I gave up. Over the past year I’ve gained a real interest in Complexity theory. I’ve read multiple books, papers, blogs… on different aspects of Complexity, Chaos, Emergence and related topics. In fact, in some of these posts I’ve referenced Complex Adaptive Systems (I’m using John Holland’s nomenclature) yet I’ve never adequately defined my terms. So, this week I decided to reread and do some additional reading to actually define my terms since, and I’m ashamed to admit this, I wasn’t completely sure what I mean by complexity. It turns out, from what I can tell, no one else is sure what the definition should be either. So, in the spirit of arrogance that comes so naturally to blogging I’ve decided to lay out some definitions below in the hope that people will want to comment and help me refine them. Actually, even if people don’t comment I’ll probably refine them. Anyway, here goes, my definitions for Complexity, Atomic Part (I might want to change that) Emergence and Complex systems.

Complexity: the the measurement of an ideal observer’s ability to predict the outcome of the interaction of a system’s atomic parts.

Atomic Parts: The arbitrarily chosen component parts that appear irreducible, when compared against each other at some arbitrarily chosen level.

Complex system: A system that can not be reduced to the sum of it’s parts. Effectively it is a system that is greater than the sum of it’s parts or as Stuart Kauffman would say, “a system that gives you something for nothing.”

Emergence: The observed qualities (or perhaps regularities) as perceived by the ideal observer that were unpredictable when said observer was given a complete description of the observed system’s atomic parts.

Caveats:

1) You’ll note in an effort to simplify my understanding of these concepts I’ve (temporarily) discarded the notion that complexity or emergence exist with out an observer. I don’t like making this assumption but I’m having trouble finding any way around it.

2) The use of an “ideal” observer is meant to do away with the argument that the unpredictability of a system is a factor of ignorance on the part of the observer. I know there are mathematical proofs that appear to prove some class of systems as complex given near infinite computational resources but I must admit I don’t have the expertise necessary to evaluate said proofs.

3) When speaking of the arbitrariness of component parts or chosen level, I’m enforcing my belief that there are in fact “levels” of a system that are internally cohesive when judging whether parts of the system are irreducible. Example, if one were to look at a human society one could just as easily argue that humans are the component parts as that human brains (or organs) or atoms are the irreducible component parts. I suspect that a part is irreducible if if possess a high enough (any?) level of complexity but, for purposes of this discussion, it seems it circular to argue this point.

4) Given the above definitions, Complexity can be an emergent property of a system, though I’d argue, that complexity will not necessarily emerge. If complexity doesn’t emerge, I’d call the Birthing System a Final Complex System and the Emergent System would be a Simple System. However, as I write this I wonder if any system is ever final since, depending on where you look, it would seem that it will birth some form of complexity even if not at a different level. Example, the interaction of ants is a complex system, the resultant ant hill is a simple system but future ant hives that come from the original ant system will, of course, be complex systems.

5) I’m still wrestling with whether complexity is a qualitative measurement, a quantitative measurement or both. In other words we can argue some system is more or less complex based on, potentially, how many complex systems it births, how many atomic parts and kinds of interactions said system starts with, how hard it is to predict Emergent phenomena…

So there it is, my first take on this issue. In my next post I’ll give some fuller descriptions of complex systems and why they’re of interest.

In a previous post I argued for the idea of regulation within an economy. Today I want to take one example from the current economic crisis, of a types of regulation I think are beneficial. First a quick overview of the economic crisis. For those interested, I’d point to some good podcasts on the issue: Fresh Air’s show with Michael Greenberger and This American Life’s show, called “Another frightening show about the economy” both of which give a more in depth discussion than this blog. And, while both shows utilize overlapping sources, I’ve found other resources echo their conclusions. With that out of the way, here we go.

The first point I want to make, the overarching point, is the ability to trade across political boundaries, the use of information technology and financial engineering has resulted in both more prosperity and more risk. We are all connected in a real time complex adaptive system which can change direction quite quickly and without any predictability.  Credit default Swaps are the nexus of the current economic crisis and highlight the advantages and disadvantages of our current global economic system.

In the late nineties companies started offering Credit Default Swaps (CDS) as a form of insurance to buyers and sellers of corp. bonds. A CDS is basically a contract wherein a buyer gets insurance from a seller against a third party’s default on a bond. Kinda like house insurance where I insure my house against burning down by paying the insurance company money each month in exchange for them promising to make me whole if disaster should strike. So far so good, but a CDS can also be used for speculation, in other words, I can buy insurance against your house burning down even though I don’t have any ownership stake in your house whatsoever. Kinda strange, I know, but here’s the thing, if I have a hunch your house is going to burn down it would be cool to make money on my hunch. But, what if I’m not 100% convinced your house will burn down? No problem, I turn around and sell a CDS to someone else, this way I’m covered. If your house burns down someone pays me and I pay someone else, if it doesn’t burn down then while I’m paying the premium on the first CDS I’ll be collecting revenue from my sale of the second CDS. If I’ve done it right, my hunch came earlier than someone else’s, my premium being paid out will cost me less than the premium I’m receiving. These latter type of transactions, where everyone “netted” their trades (both a buyer and a seller of risk) was the rule rather than the exception. In fact, of the $5 Trillion in corp. debt there is an estimated $62 Trillion in CDSs. In other words, for every $1 of corp debt there is approximately $10 of speculation. That’s what’s called leverage and it’s all interconnected. As you can guess, if one party fails to pay off their obligation it will ripple and multiply throughout the system. On the other hand, if leverage works, it means credit is easier to get, more things are built, more investment is made, more jobs are created…

In 1998 regulators started arguing the CDS market should be regulated. The counter argument, made by both the Clinton administration as well as Phil Gramm Republicans, was the buyers and sellers of these products are sophisticated consumers, and, as such, let the free market work. In 2000 Phil Gramm slipped an amendment into an omnibus bill making it law that CDSs would not be regulated. As I understand it, not one senator (Democrat or Republican) opposed the amendment. But here’s the rub. The economy is situated within a larger interconnected system. If there is a disruption because of these financial instruments then a large number of “innocent bystanders”, people who had no part of the transaction, will be harmed. Meaning, there’s a role for government to at least regulate this market enough to know if harm is coming and to minimize some of the chances for harm.

Not allowing these instruments is draconian and counterproductive. Insurance has a long history of being useful in capitalist economies, it encourages risk taking and provides people with a partial shield against bad luck. Even the kind of speculation involved in CDSs where the owners of the bond were not involved has it’s advantages. It encourages the spreading of information and the interconnection of agents that allow for more leverage.

However, historically insurance as an industry has been regulated. Companies have to have a certain amount of capital reserved in case they are forced to pay off insurance policies. Further, there is a certain amount of transparency in the transactions so other players can adequately gauge risk and reward. These are the kind of regulations that should have been in place for the CDS market. As written right now, a CDS is a private contract between two consenting parties. There is no way for anyone to see how much was paid for a CDS or even how many players own or sold CDSs. Further, there is no mechanism to make sure the sellers of a CDS are adequately capitalized.

These last three points are key, without a transparent pricing system, if there’s a shock to the system (like the sub-prime crisis) there’s a lack of trust in the marketplace. No one knows who’s balance sheet is real, further due to the interconnection of all these transactions a bank failure in Indonesia can hurt an American hedge fund six steps removed. The lack of a central clearinghouse means no one knows where these, as Buffett called them, “financial weapons of mass destruction” are located. The govt. has no idea how bad things can or will get and neither does anyone else. Finally, because CDSs were not regulated there was no way of making sure a seller was adequately capitalized and it encouraged more players into the market thus increasing the likelihood there would be this kind of a crisis.

Which brings me back to regulation. In a complex adaptive system, like an economy, regulation must allow for changes, growth and the death of individual agents but minimize the risks of the whole system suddenly going through a radical reorganization. Not because radical reorganizations are neccesserily bad in and of themselves but because humans and human society needs time to react and adjust. Further, regulation should minimize some of the more extremes of group psychology (like panic) while also combatting harmful concentrations of power and knowledge.  In retrospect, we can only wonder if things in the CDS market would be as bad if pricing was transparent, purchases were registered with a central clearinghouse and sellers were forced to be adequately capatilized.

If you ever see three ducks sleeping on a log, watch them for a bit. The duck in the middle will comfortably sleep, pretty much staying in one position. The two ducks on the outside will sleep with the eye closest to the middle closed and the eye on the outside open then, after a little while, the ducks on the outside will stand up and turn 180 degrees, close the eye that was open and open the eye that was closed. Only half of the duck’s brain is sleeping at one time. Other birds, reptiles, insects, sea mammals… do this as well, (it’s called logging in dolphins because they look like logs floating in the water). The animals sleep half a brain at a time (uni-hemispheric sleep) so the other half can watch for enemies (Dolphins also need to make sure they continue breathing since they’re conscious breathers, they have to come to the surface to breath). Uni-hemispheric sleeping raises a couple of questions.

1) How do you get to be the duck in the middle? Seems like it’s a lot more restful.

2) How do ducks, dolphins and other uni-hemispheric sleepers solidify long term memories whose neuronal maps span both hemispheres?

The current working assumptions, as I understand them, is that sleep is universal (every animal from insect to mammal sleeps or does something that looks like sleep) and sleep seems to serve multiple purposes; cleansing of neurotransmitters to allow more learning the next day, REM sleep allowing the brain to consolidate what is learned during the day and mechanisms during REM and Non-REM sleep for forgetting the non-useful data. If something is learned it will be represented by a “neural map”, effectively a grouping of neurons that may span different areas of the brain, including both hemispheres. During sleep this grouping (since it has been deemed valuable) will be strengthened. For example, if a rat is run through a maze during the day and is rewarded with food, then, during sleep, it will dream it’s running through the maze in order to consolidate its learning on how to get the reward. What confuses me, is how is this learning effected by uni-hemispheric sleeping? In other words, if a given neuronal grouping crosses hemispheres is learning consolidation improved, lessened or is there no difference?

This post was inspired by a podcast by one of my favorite radio shows Radiolab. Another of my favorite shows, This American Life did a less scientific, though quite entertaining look at sleep.

In the previous two posts I’ve discussed neuroplasticity as a concept and Schwartz’s methodology for treating Obsessive Compulsive Disorder (OCD). In this post I’ll look at the five steps to Schwartz’s underlying theory:

1) The OCD circuit locks up, firing the same message over and over. Schwartz doesn’t provide any firm answers why this happens but he’s supported by a good amount of science when arguing the “locking” metaphor is reasonably accurate.

2) The therapist encourages the patient to come up with an alternate idea, for example, rather than washing their hands they could go work in the garden. This idea activates the planning circuits in the prefrontal cortex. Initially this circuit will be considerably weaker than the circuit represented by the OCD enabled idea. As mentioned in the first post the relative strength of the circuits is partially determined by which circuit has fired more and, perhaps more importantly, which circuit has received more attention.

It is the next three steps that represent a radical departure from standard neuropsychiatry.

3) Here Schwartz focuses on the nerve terminal of the neurons in both the “wash hands” as well as “go to the garden” circuits. He points out the size of this part of the neurons (the ion channel) where neuro-transmitters cross, activating or suppressing the target neuron’s receptors is so small, its behavior may be determined by Quantum Mechanic rules. Schwartz argues the vesicle which will either release or not release the neurotransmitter is in a superposition of quantum wave function. Meaning, rather than being solidly in one state or another (release or don’t release), it’s actually in a probability state, both releasing and not releasing at the same time. The cat is both dead and alive.  Now, while it’s true the scale of these structures are small enough to be subject to QM rules, it’s not been proven they are subject to these rules.

4) While the Quantum rules allow for “release” and “don’t release” to co-exist, it’s also true the state representing the OCD compulsion (wash your hands) has a higher probability of being true. This is where Schwartz’s treatment comes into play.

5) By expanding mental effort, focusing their attention on the healthier alternative, the patient utilizes  (in Schwartz’s words) mental force to change the odds and increase the probability of that circuit being activated. This works because of the “Quantum Zeno effect”. Quantum Zeno is an effect, recognized by physicists, that roughly states, if an observer rapidly and repeatedly observes an event, those series of observations can freeze the properties of that event into one state rather than allowing it to change back into a wave function. In effect, Schwartz is arguing, by concentrating on the circuit the patient wants to activate s/he can increase the odds the circuit will be activated. If done enough and with enough attention that circuit will eventually extinguish the OCD circuit.

Schwartz can back up his theories with fMRI and PET scans showing the physical changes to the brain solely based on the patent’s use of these cognitive strategies. In this case it’s the mind that changes the brain, not the other way around. For a scientific materialist this presents quite a problem. If everything comes from matter and only matter can change matter than how is matter changed by simple thought? The ramifications of this theory provide an interesting argument in opposition to reductionist thinking. Perhaps there is not just a brain that gives rise to a sense of a mind, but rather something that exists beyond or in conjunction with the brain. A mind that can change the physical structures of a brain just as the brain can change the mental structures of the mind.

While I’m not knowledgable enough to question all the assumptions underlying this theory I do find it an interesting counterpoint to the more traditional scientific materialism. Needless to say, my three blogs barely do justice to Schwartz and Begley’s book, but hopefully, they’ll encourage you to read the book yourself.