Dear Bob, 

I meant your book "The Third Window", but apologize for having been not so
clear. I reacted to Bob Logan's review of it that he distributed on this
list. The description of autocatalysis in the case of three reactants
clarifies your notion of "ascendency" that remained previously a bit hidden
because of the biological lingo. Perhaps, you have also taken it further. 

Unfortunately, the figure that you discuss did not make it in the
transmission. Perhaps, you can resend.

> Somewhere I also remarked that a negative mutual information might act
like the Pauli-Exclusion principle in physics. 
> That is, it would signify an untenable or unstable configuration. 

Whereas mutual information among two dimensions can stabilize as in "mutual
shaping" or "action = - reaction", three dimensions can also lead to
meta-stabilization, hyper-stabilization, and also globalization. In your
case of systems formation because of autocatalysis, for example, the system
is globalized among the three carrying systems in the sense that it is not
localized with one of them or one specific link between two of them. In this
sense, the system remains in transition. 

Best,
Loet

-----Original Message-----
From: Robert Ulanowicz [mailto:[email protected]] 
Sent: Monday, February 25, 2013 7:34 PM
To: Loet Leydesdorff
Cc: 'Robert Ulanowicz'; [email protected]
Subject: RE: [Fis] Cognitive Effects of Cognitive Research: Photographic
evidence

Quoting Loet Leydesdorff <[email protected]>:

> Dear Bob,
>
> I now read the book: the description of autocatalysis is very 
> convincing. It really clarifies how a "triple helix" (of 
> university-industry-government
> relations) can operate in generating synergy (reducing uncertainty).

Dear Loet,

I'm unsure to which book you are referring? "A Third Window" or "Ecology,
the Ascendent Perspective"?

> You then go on with "Average Mutual Information" (AMI). You don't 
> mention that AMI (that most of us call "mutual information") is 
> necessary positive since a Shannon-type information, but that "mutual 
> information in more than two dimensions" can be negative and thus used 
> as an indicator of synergy or autocatalysis.

It's true, I don't refer to the possibility of negative 3-D mutual
information in any of my books. What Claudia Pahl-Wostl and I agreed to call
information in the 3-D case was the entire ensemble (1+2+3+4) in the
attached diagram. While 1 can go negative, the full ensemble does not.

Recall that information is always relative, so whether a particular
component is viewed as information or as conditional entropy is a matter of
perspective. For example, any of the "ears" 2, 3, or 4 can be viewed as a
conditional entropy in the context of all three variables, or they can be
viewed as information in the context of their respective binary
associations.

Finally, Gennaro Auletta in his book, "Cognitive Biology: Dealing with
Information from Bacteria to Minds", maintains that complexity is a property
of more than 2 dimensions and that the magnitude of the "ears" is a gauge of
such complexity. His definition makes some sense in that an "ear" represents
how two variables influence one another in complete abstraction of variation
in the third -- certainly one manifestation of complex behavior.

> I usually make reference for this to Ulanowicz (1986, pp. 143 ff.). Is 
> that methodologically the same argument? I assume so. Or have you 
> taken these measurement issues also further?

That reference is fine. I really haven't taken this issue much further. I do
remark on p336 of <http://people.biology.ufl.edu/ulan/pubs/METHODS.PDF> that
the 3-D or higher mutual informations can be negative. Somewhere I also
remarked that a negative mutual information might act like the
Pauli-Exclusion principle in physics. That is, it would signify an untenable
or unstable configuration. Unfortunately, I can't find where I wrote that
speculation. Sorry.

> Best,
> Loet

The best as always,
Bob

_______________________________________________
fis mailing list
[email protected]
https://webmail.unizar.es/cgi-bin/mailman/listinfo/fis

Reply via email to