Re: models, sensory perception

[Howard Pattee <***>]

Tim,

I will try to explain why I brought up what I found  ambiguous in the your concept of the 
NP. Let me make clear that I am in complete agreement with Rosen that NP is not a 
universal format. What I disagree about is the claim that NP fairly characterizes all of 
modern physics and that physicists themselves believe that NP is universal. In other 
words, there is no disagreement here except a misconception about physics.

In LI, p. 134, Rosen says about relational models: ?As I have developed it so far, there 
is no time parameter, no states, no state transition sequences. There are only components 
(mappings), and organizations, the abstract block diagrams that can be built from them.?

I asked about the extremum (variational) and symmetry principles just because they also 
have no time parameter, no states or state transition functions. These are called 
principles, but they are a type of relational model. These types of models one could call 
synchronic because time has been abstracted away. In any case, they are often considered 
more foundational in physics than any particular dynamics.

I wondered whether you would recognize them as a type of relational model. The quote you 
gave (EL p. 216) is ambiguous. Rosen says variational principles seem to ?violate? 
state-determination, and this ?has always bothered people,? but he does not pursue the 
issue. It has bothered some philosophers, but it does not bother physicists. They regard 
it as a useful complementary model. It apparently bothered Rosen because the principle is 
not within the NP. Symmetry principles, each of which leads to a conservation law 
(Noether?s theorem), are even more fundamental in modern physics and they certainly are 
not part of the NP.

Rosen and I have always agreed that there is no complete model of anything. In fact, one 
of his measures of system complexity was the number of models needed to answer basic 
questions about the system. Abstract relational models are fundamental to modern physics, 
but by themselves synchronic models are not directly testable models. Physics complements 
timeless models with time-dependent models, time-symmetric models with time-antisymmetric 
models, and discrete particle models with continuous field models. These models may even 
appear inconsistent if one tries to combine them formally into a single model. This is 
all outside the NP.

We agree that in science we must make empirical contact with reality somewhere. What this 
means is that there must be measurements on states. That is all physicists (and Rosen) 
mean by a state. States can be simple or complex, defined by numbers or just patterns. 
But numbers and patterns must be ultimately be recognized locally in ordinary classical 
space and time. This is just as true for variational and symmetry models as for other 
relational models. It holds for cells, evolution, QM, and all physical models. If string 
theory does not eventually stimulate a model with a measurable consequence it will die 
out.

Rosen defines states this way in AS. Following the modeling diagram on p. 74, he says: 
?(italics) a state embodies that information about a natural system which must be encoded 
[by observation and measurement] in order for some kind of prediction to be made. It 
should be explicitly noted that, according to this definition, the concept of state of a 
natural system is not meaningful apart from a specific encoding of it into some formal 
system.? In other words, a state is defined only by what you decide to measure. He 
elaborates this on p. 127 where he also notes the necessity of the epistemic cut  (Fig. 
3.2.1). There, Rosen says, ?The dotted line (in the Fig.) bisecting the diagram separates 
those parts which pertain to the external world from those that pertain to the world of 
formal systems.?

So I think everyone agrees that formal state-determined, single-time dynamics is neither 
universal nor a lingua franca. This issue is a straw man. Rosen only in LI, not in AS, 
sounds like physicists have somehow imposed this artificial restriction to the NP on 
themselves. This is certainly not how physicists think. In fact, this is just reversing 
cause and effect. It is true that many universal laws of nature can be expressed by a 
memoryless, state-determined dynamics. But this is not because physicists imposed this on 
nature by restricting themselves to NP. This is because it is the nature of simple 
particles to have no internal memory. Furthermore, the conceptual basis from which these 
laws were derived is clearly outside the NP. The formalization of the laws is only the 
last step in a long process of nonformalizable, non NP conceptual models. Many 
solid-state models are not state determined.  Protein folding models are a good 
biological example.

Rosen recognized all this in AS, but in LI he tries to support his own views by 
criticizing physics. Rosen?s relational views must stand on their own, and in fact 
physicists are much more in tune with Rosen?s relational models than he appreciated. As I 
said earlier, introducing relational concepts into the biological community is enough of 
a problem without antagonizing physicists. Leave out the umbrella complaints about 
physics. What Rosen modelers should work on are biological models based on his relational 
principles that have good observables that can define states.

Howard