Dear Terry and colleagues,
Thanks a lot for the opening text! It is a well crafted Essay full of
very detailed contents. My impression is that the "microphysics" of
information has been solved elegantly --at least at the level of today's
relevant knowledge-- with your work and the works of related authors,
one of them Karl Friston, who could be linked as a complementary
approach to yours (in particular his recent "Life as we know it", Royal
Society Interface Journal, 10: 20130475). His Bayesian approach to
life's organization, coupled with (variational) "free energy"
minimization principle, conduces to the emergence of homeostasis and a
simple form of autopoiesis, as well as the organization of
perception/action later on. Thus, quite close to your approach on
autogenic systems. About the different sections of the Essay, the very
detailed points you deal with in section 4 ("steps to a formalization of
reference") are, in my opinion, the conceptual core and deserve a
careful inspection, far more than these rushed comments. In any case,
the relationship Boltzmann-Shannon entropies has been cleared quite
elegantly.
However, for my taste the following sections have not sufficiently
opened the panorama. And with this I start some critical appreciations.
Perhaps the microphysics of information is not the critical stumbling
block to me removed for the advancement of the informational
perspective. We could remain McLuhan's stance on Shannon's information
theory and von Neumann's game theory... yes, undoubtedly important
advancements, but not the essential stuff of information. But in this
list there are people far more versed in McLuhan's contents and whether
the caveats he raised would continue to apply (obviously in a different
way). I am also critical with the autogenesis model systems--wouldn't it
be far clearer approaching a (relatively) simple prokaryotic cell and
discuss upon its intertwining of the communication and self-production
arrangements? The way a bacterium "sees" the world, and reorganizes its
living, could be a very useful analysis. I think it leads to a slightly
different outcome regarding reference/significance, and
meaning/value/fitness.
If we look at the whole view of the "information world" (human
societies, behaving individuals, brain organization, cellular processes,
biomolecules) and how a myriad of information flows are crisscrossing,
ascending, descending, focusing, mixing and controlling energy flows,
etc. we may have an inkling that this evanescent world paradoxically
becomes the master of the physical world (the "fluff" versus the
"stuff", Lanham 2006), and that is organized far beyond the rules of the
micro-macro-physical world. But how? What are the essentials of this
magnificent "castle in the air" (reminding Escher's engrave:
http://fis.sciforum.net/ )?
In next exchanges I will try to ad some more specifics on the above
"fluffy" comments, derided from a fast reading of the Essay. Thanks
again, Terry, for providing us this discussion opportunity in the New Year.
best ---Pedro
_*Steps to a theory of reference & significance in information
*_*FIS discussion paper by Terrence W. Deacon (2015)*
This is the link to download the whole paper:
https://www.dropbox.com/s/v5o8pwx3ggmmmnb/FIS%20Deacon%20on%20information%20v2.pdf?dl=0
/"The mere fact that the same mathematical expression - Σ pi log pi
occurs both in statistical
mechanics and in information theory does not in itself establish any
connection between these
fields. This can be done only by finding new viewpoints from which
thermodynamic entropy and
information-theory entropy appear as the same concept." /(Jaynes 1957,
p. 621)
/"What I have tried to do is to turn information theory upside down to
make what the
engineers call 'redundancy' [coding syntax ] but I call 'pattern' into
the primary
phenomenon. . . . “/ (Gregory Bateson, letter to John Lilly on his
dolphin research, 10/05/1968)
*Introduction*
In common use and in its etymology the term ‘information’ has always
been associated with
concepts of reference and significance—that is to say it is about
something for some use. But
following the landmark paper by Claude Shannon in 1948 (and later
developments by Wiener,
Kolmogorov, and others) the technical use of the term became almost
entirely restricted to refer
to signal properties of a communication medium irrespective of
reference or use. In the
introduction to this seminal report, Shannon points out that although
communications often have
meaning, “These semantic aspects of communication are irrelevant to
the engineering problem”
which is to provide a precise engineering tool to assess the
computational and physical demands
of the transmission, storage, and encryption of communications in all
forms.
The theory provided a way to precisely measure these properties as
well as to determine
limits on compression, encryption, and error correction. By a sort of
metonymic shorthand this
quantity (measured in bits) came to be considered synonymous with the
meaning of
‘information’ (both in the technical literature and in colloquial use
in the IT world) but at the cost
of inconsistency with its most distinctive defining attributes.
This definition was, however, consistent with a tacit metaphysical
principle assumed in the
contemporary natural sciences: the assertion that only material and
energetic properties can be
assigned causal power and that appeals to teleological explanations
are illegitimate. This
methodological framework recognizes that teleological explanations
merely assign a locus of
cause but fail to provide any mechanism, and so they effectively mark
a point where explanation
ceases. But this stance does not also entail a denial of the reality
of teleological forms of
causality nor does it require that they can be entirely reduced to
intrinsic material and energetic
properties.
Reference and significance are both implicitly teleological concepts
in the sense that they
require an interpretive context (i.e. a point of view) and are not
intrinsic to any specific physical
substrate (e.g. in the way that mass and charge are). By abstracting
the technical definition of
information away from these extrinsic properties Shannon provided a
concept of information that
could be used to measure a formal property that is inherent in all
physical phenomena: their
organization. Because of its minimalism, this conception of
information became a precise and
widely applicable analytic tool that has fueled advances in many
fields, from fundamental
physics to genetics to computation. But this strength has also has
undermined its usefulness in
fields distinguished by the need to explain the non-intrinsic
properties associated with
information. This has limited its value for organismal biology where
function is fundamental, for
the cognitive sciences where representation is a central issue, and
for the social sciences where
normative assessment seem unavoidable. So this technical redefinition
of information has been
both a virtue and a limitation.
The central goal of this essay is to demonstrate that the previously
set aside (and presumed
nonphysical) properties of reference and significance (i.e.
normativity) can be re-incorporated
into a rigorous formal analysis of information that is suitable for
use in both the physical (e.g.
quantum theory, cosmology, computation theory) and semiotic sciences
(e.g. biology, cognitive
science, economics). This analysis will build on Shannon’s
formalization of information, but will
extend it to explicitly model its link to the statistical and
thermodynamic properties of its
physical context and to the physical work of interpreting it. It is
argued that an accurate analysis
of the non-intrinsic attributes that distinguish information from mere
physical differences is not
only feasible, but necessary to account for its distinctive form of
causal efficacy.
Initial qualitative and conceptual steps toward this augmentation of
information theory have
been outlined in a number of recent works (Deacon 2007, 2008, 2010,
2012; Deacon &
Koutroufinis, 2012; Deacon , Bacigaluppi & Srivastava, 2014). In these
studies we hypothesize
that both a determination of reference and a measure of significance
or functional value can be
formulated in terms of how the extrinsic physical modification of an
information bearing
medium affects the dynamics of an interpreting system that exhibits
intrinsically end-directed
and self-preserving properties.
[...]
A model system
To test these principles and their relationship to reference and
significance, I and my
colleagues have conceived of an empirically realizable and testable
thought experiment. As in
most efforts to formalize basic physical properties it is useful to
begin with a simple model
system in which all aspects of the process can be unambiguously
represented. For our purposes
we describe a theoretical molecular system called an autogen, which
maintains itself against
degradation by reconstituting damaged components and reconstituting
system integrity. This
model system involves an empirically realizable molecular complex
described previously
(Deacon 2012; also in Deacon & Cashman 2012; and also called an
autocell in Deacon 2006a,
2007; 2009; and Deacon & Sherman 2008).
[...]
In this way we can use formal and simulated versions of autogenesis to
develop a measure of
relative significance, in the form of “work saved.” I hypothesize that
this simple model system
exemplifies the most basic dynamical system upon which a formal
analysis of informational
interpretation and significance can be based.
[...]
In both forms, modifications of the autogenic process is provided with
information referring
to its own preservation via boundary conditions (external or internal)
that are predictive of
successful self-preservation. The significance of information of
either sort is assessed by the
relative minimization of work per work cycle, and therefore the
decreased uncertainty of selfreferential
constraint preservation. In this way interpretation is analogous to
the decrease in
uncertainty that is a measure of received information in Shannonian
theory, but at a teleodynamic
system level.
Using these three variants of a simple model system I claim that we
can precisely analyze the
relationships between information medium properties, intrinsically
end-directed work, and the
way these enable system-extrinsic physical conditions to become
referential information
significant to system ends. These relationships are not only simple
enough to formalize, but they
can be simulated by computer algorithms at various levels of logical
and physical detail. I
believe that creating and experimenting with these simulated autogenic
systems will enable us to
reframe the mysteries of reference and significance as tractable
problems, susceptible to exact
formal and empirical analysis. This is still a far cry from a theory
of information that is
sufficiently developed to provide a basis for a scientific semiotic
theory much less than an
explanation of how human brains interpret information, but it may
offer a rigorous physical
foundation upon which these more complex theories can be developed.
*— Terry*
Professor Terrence W. Deacon
University of California, Berkeley
-------------------------------------------------
Pedro C. Marijuán
Grupo de Bioinformación / Bioinformation Group
Instituto Aragonés de Ciencias de la Salud
Centro de Investigación Biomédica de Aragón (CIBA)
Avda. San Juan Bosco, 13, planta X
50009 Zaragoza, Spain
Tfno. +34 976 71 3526 (& 6818)
pcmarijuan.i...@aragon.es
http://sites.google.com/site/pedrocmarijuan/
-------------------------------------------------
_______________________________________________
Fis mailing list
Fis@listas.unizar.es
http://listas.unizar.es/cgi-bin/mailman/listinfo/fis