I wish to remark briefly on SA's remarks concerning supervenience, and then on his comment regarding my own contribution.

I concur with his initial three points (the definition in logical terms, its unclarity, its being vacuous). I would only like to add that the debates in the literature are a lot more interesting and readable than comparable debates raised in the past in terms of analytic philosophy. For lurkers who have not had the opportunity to explore the issue, it is probably worth doing. Jaegwon Kim, for example, writes clearly and persuasively (although his positions shifts). That is, while I find little use for the concept, exploring it is a nice little exercise.  What I draw from the discussion of supervenience is a shared intuition that it is necessary to define the base level (or initial state) in such terms that it allows for the emergence of novelty, but there is no consensus how we should go about doing it. Some have proposed that (unobservable) functional properties be included in the description of the base level, but there seem persuave reasons why functional properties are problematic or beg the question. 

As regards my last post, my aim is to replace the assumption that all explanation is causal with explanation in terms of a relation of processes. However, not "process" understood as a sequence of states of affairs or events, for this reduces to a causal explanation. The basic difference between a causal explanation and my processual explanation that the classic causal explanation lends to the cause a capacity to produce the effect, or the base level has the capacity to rive rise to the emergent properties of the outcome or whole, or has something (such as a mark or energy) that is transmitted to the emergent level. What I try to do is to define all levels as being themselves an emergent process, and so a relation of base to emergent level or intitial state of affairs to its outcome is understood as an actualization of a possibility that belonging to the emergent level or outcome, and the initial state or base level only qualifies that result, rather than produce or generate it. A simply way to put this is that causal explanation sees the cause as active agent and the effect as a passive recipient. I see both as active processes. More accurately, I see both "cause" and "effect" as two constraints upon a common shared process (universal dissipation) that in this case have entered into communication.

Now this is rather abstract, but when I try to pin it down, it gets a little hairy. First, it would require a definition of "process". For better or worse, I would define process as the constraint of structure (the past) upon exogenous possibilities (the future possibilities of the more universal level on which it is a constraint), and as driven by an extrinsic probability gradient (the probability differential between an actual and a possible state of affairs in the present). Together, the modalities of actuality (structure), possibility (the future) and potency (motion in the present) are a probability distribution and are how we can represent a probability distribution in thought. Note that assumed here is a scientific realism, a modal realism (not Lewis' plurality of worlds), and a version of presentism.

But, still, just what goes on when two processes enter communication remains unspecified. A fairly uncontentious place to start is to suggest that when two processes (probability distributions) enter into communion, they frame each other, which is to say they localize each other to give possible properties actual values. This is a big subject much discussed and so I won't elaborate on it here. But it still does not go very far into defining just happens in what we conventionally represent in thought as a causal relation.

Second, in conventional causal explanation, the cause dissipates energy as it does work and produces a change in the affected entity. Instead, I see this dissipation as changing the probability distribution of the other process by making one of its hitherto improbable outcomes a) probable and b) actual. The structure of the "causal" process, within the constraint implied by the structure of the affected process (multiplication of probability values) is combined with the probability gradient of the "causal" process to make this improbable possibility very probable. While this way of seeing things puts a little strain on the brain, I think it really describes pretty well what we know is going on.

Note that I've simplified things here for the sake of explanation. The logic of the conception I propose implies that there are no causes and effects, for all processes in communication are both. I suggest that what distinguishes the cause is merely its having a greater probability gradient than the effect, but in principle what is happening in each process is the same. Our sense of temporal assymetry is therefore the effect of the accident of which process has the greatest probability gradient.

Now S.A. brings up non-linear feedback, for which I suppose we can thank Prigogine. However, I'm not comfortable with it. For one thing, non-linearity is used to account for non-equilibrium systems, and how this might relate to "causality" in equililbrium systems or the relation of an emergent level to its base level is obscure. But even on Prigogine's turf, while non-linearity might be adequately descriptive, I'm not sure it is explanatory. That is, non-linearity describes emergence (in the old definition) rather than explain it. However, I defer to those better informed than I do deal with this.    

I don't know that discussions of supervenience imply relations are "predetermined", for it seems to me that they often rely on a black box. However, the word predetermined or determined can be understood in different ways. If the emphasis is on "pre", then would agree to the objection to supervenience. If the emphasis is on "determined", then we must decide between unequivocal (Laplacean) determination, and of course no one buys that, but I don't think the supervenience folks do either. If the determination is probabilistic, I suspect most folks would find it agreeable, but it seems to beg the question as description rather than expose the mechanism of this probabilism.

Incidentally, I was interested in SA's citations. Unfortunately the Teller book is not in Amazon, and I couldn't find the Bailey article in JStor. I'm trying to remember if JStor covers Synthèse. Any idea what my problems could be here?

Haines Brown