Sunday, 4 November 2007

Causation is a cosy state of mind

To assume causation as a possible explanatory avenue is a cosy state of mind indeed, and it is in the interest of existing structures to maintain such cosiness. The variety of perspectives through which a situation can be observed leads one progressively to replace objective causation with other forms of connection. Thus, attribution, parallel distance, asymmetry etc are ways that currently float about, with the important feature of being uniquely incapable of replacing every other perspective and offer an adequate description of unity.

In such a draughty attributional format, where one observes causation without the possibility of conclusive proof, the one thing that international politics can do is abandon the theorisation of a hierarchical causal structure, its parts connected through direct control or even influence, and accept the multiplicity of differentiated realities. This can only mean that, however close one looks into causation, one can only come up with an attribution and this is all right. A reconceptualisation of distance, both in terms of causation and in terms of physicality, as the necessary enabler of development (in the evolutionary rather than the growth-targeting sense) is becoming increasingly more apparent in the international financial markets, where unpredictability (in the sense of perspectival cognitive isolation) is precisely the driving force behind the market itself.

What is more worrying is the public attribution of consequences to such a drive. In other words, and to put it in a slightly (certainly not radically) different parlance, does legitimacy need causation? And if so, is legitimacy a cosy state of mind too?


John Flood said...

This is tricky. Our social world is such that causation will always be a gloss on a series of strong correlations and associations. Whereas in the physical world causation may be more direct. Yet we must deal with it. Is Musharraf's new military coup caused by the Pakistani Supreme Court's expected decision that Musharraf cannot continue to be both commander in chief and president? Most likely yes. From his perspective it's a preventative measure; from ours it's the abnegation of democracy. Now this does raise questions of legitimacy and I would consider it important that some form of correct causation is correctly assigned. Perspective is important. And I wouldn't like the Pakistani government in collusion with the US and UK governments to be able to pass this off as a mere hiccup on the road to democracy. That would be cosy. While it may not be the only cause, it is a strong candidate.

joetanega said...

I suppose a curious thing about causation, a state of mind and structures is their abstractness begs for unity. This is hardly cozy. We glimpse affinities and similarities in their wide connotations. To pin a theory which stretches across physical causation and the whirling Machiavellianism of international war and politics would be ambitious. But given the dangerousness of the complex leveraged systems of financial markets and nuclear-tipped governments, it's a challenge that's worth a real try. Morgenstein & Von Neumann tried with their Theory of Games and Economic Behaviour (1944) which seeded cold-war games of reciprocal brinkmanship in the 1950s and 60s. In any case, failure in the theoretical would only be superseded by catastrophic failure in the physical. The following is an extract from an unpublished paper on "Risk Symmetries" that examines a few basic ideas about causation across the physical-social divide:

1.1 Causation, Correlation and Mutual Information Communication Systems

Causation and symmetry arguments are closely linked. What is the cause of an event A? The answer of course lies somewhere in the region of not-A. It is interesting to note that the theory of statistics itself is based on the concept that correlation itself is symmetric. Formally, “for any two events A and B are said to be correlated if fixing the outcome of B can change the probabilities for the outcome of A. That is, if p(a|b)  p(a) for some a,b. Thus p(a|b)  p(a) if and only if p(b|a)  p(b).” There are perhaps millions of words spilled on the problem of defining the relationship between A and the cause of A, and the precision and accuracy of the definitional relationship are at the core of how we define the rules of rules (the meta-rules) of thought. Causation itself appears to set out an asymmetry. However, the let’s say A causes B. Using Seth Lloyd’s (1992) directional script, A -> B. This appears to be a non-symmetric or asymmetric relationship. Leibniz’s principle of sufficient reason, for example, states that all things being equal, nothing changes without there being a sufficient cause. But our contention is that again, at the level of a meta-explanation of an uni-directional cause, is a symmetric model. Classical problems of causation are simply aspects of underlying symmetry arguments. Let us explore how the asymmetry of causation can be explained in terms of symmetry.

Seth Lloyd (1996) has shown a logical symmetry exists in our fundamental concepts of causation based on correlation, and that this symmetry can be extended through information theory (Shannon and Weaver 1949) to specify mutual information between systems, which is also at its heart, symmetric.
The basic idea here is that causal connection implies the possibility of statistical correlation, while the absence of causal connection implies the absence of correlation. If two events are correlated, then either one has an effect on the other, or the two have a common cause in their past. By extending these ideas to many events, one arrives at a method for deriving patterns of statistical dependence and independence from the causal relations between the events. [I'm working on a group theoretic solution, for example, to judicial reasoning.]

From correlation symmetry, we also see that information between communicating systems is symmetric. Suppose we have two communication-information processing systems, Alpha and Beta. Initially from either Alpha’s or Beta’s perspective, the initial message from the other may appear undecipherable, undetermined or chaotic. However, as Alpha and Beta interact, there will be a mutuality of information, that is, an exchange of messages. The message may be defined as the perceived asymmetric difference in information from the perspective of one party. However, between two parties, the information of the message is exactly the same. Thus, whilst each party perceives a decrease in perceived randomness upon accepting a message, there is simply mutual information that affects both parties. This information asymmetry seen from one party and simultaneous information symmetry from both parties, means that both parties can exchange a message (a bit of surprise) that is immediately incorporated into their own respective systems. The reduction of complexity or randomness from either system’s perspective when viewed from a higher level where both systems are part of one larger interacting system, is simply mutual. One system is not more complex than another. Rather, both have logical depths that are infinitely capacious. As stated in Wolfram’s Principle of Computational Equivalence, “almost all processes that are not obviously simple can be viewed as computations of equivalent sophistication.”

1.2 Application to Luhman’s Systems Theory

This line of thought may be applied to legal theory. Consider Luhmann’s approach to defining the legal system, asserting that law is based on an “autopoietic”, “autonomous” and “operational closed” system. Luhmann’s description of the law as a “self-reproducing network which relies exclusively on self-generated information and is capable of distinguishing internal needs from what it sees as environmental problems” is simply one half of a necessary symmetric relationship with the environment. As Luhman states the problem of defining the law,
As part of the societal system, the legal system is a self-organizing, self-determining system….As in all cases of operational closure, the problem is how to define the operation that differentiates the system and organizes the difference between system and environment while maintaining reciprocity between dependence and independence. This indeed is a most difficult question; it is the core problem of a theory of the legal system.

According to Luhmann, that which defines the boundary of the legal system is the “binary code” “that is, the continuous necessity of deciding between legal right and wrong”. (Note here the invocation to logical completeness.) This strict distinction, reified to the notion of “boundary” between the law and its environment is kept intact by the operation of the binary code within the legal system. Luhmann further argues that the “bifurcation necessitates decisions and thereby further operations, and decisions require the construction of normative rules (programs) to connect them in a network for reproducing decisions.” The point is that Luhmann’s description of the legal system is not as it appears to argue for the special case of the legal system, but rather that his argument is a description of only one side of the “mutual information” symmetry.