Provided by Cognitive Sciences ePrint Archive
Version 16
GENE EXPRESSION AND ITS DISCONTENTS
Developmental disorders as dysfunctions of epigenetic cognition
Rodrick Wallace, Ph.D.
Division of Epidemiology
The New York State Psychiatric Institute*
March 30, 2009
Abstract
Reductionist treatments of the epigenetic regulation of gene
expression suffer the same mereological and sufficiency falla-
cies that haunt both contemporary systems biology and neu-
ral network models of high order cognition. Shifting perspec-
tive from the massively parallel space of gene matrix inter-
actions to the grammar/syntax of the time series of devel-
opmentally expressed phenotypes using a cognitive paradigm
permits import of techniques from statistical physics via the
homology between information source uncertainty and free en-
ergy density. This produces a broad spectrum of ‘coevolution-
ary’ probability models of development and its pathologies in
which epigenetic regulation and the effects of embedding en-
vironment are analogous to a tunable enzyme catalyst. A
cognitive paradigm naturally incorporates memory, leading
directly to models of epigenetic inheritance, as affected by
environmental exposures, in the largest sense. Understanding
gene expression, development, and their dysfunctions will re-
quire data analysis tools considerably more sophisticated than
the present crop of simplistic models abducted from neural
network studies or stochastic chemical reaction theory.
Key Words developmental disorder, epigenetic cogni-
tion, gene expression, information theory, merological fallacy,
phase transition
1 Introduction
1.1 Toward new tools
Some researchers have recently begun to explore a de-facto
cognitive paradigm for gene expression in which contextual
factors determine behavior of what Cohen calls a ‘reactive sys-
tem’, not at all a deterministic, or even stochastic, mechanical
process, (e.g., Cohen, 2006; Cohen and Harel, 2007; Wallace
and Wallace, 2008). This approach follows much in the spirit
of the pioneering efforts of Maturana and Varela (1980, 1992)
* Address correspondence to: Rodrick Wallace, 549 W. 123
St., Suite 16F, New York, NY, 10027 USA, 212-865-4766, wal-
[email protected].
who foresaw the essential role that cognitive process must play
across a broad spectrum of biological phenomena.
O’Nuallain (2008) has, in fact, placed gene expression
firmly in the realm of complex linguistic behavior, for which
context imposes meaning, claiming that the analogy between
gene expression and language production is useful, both as a
fruitful research paradigm and also, given the relative lack of
success of natural language processing (nlp) by computer, as a
cautionary tale for molecular biology. In particular, he states,
given our concern with the Human Genome Project (HGP)
and human health, it is noticeable that only 2% of diseases
can be traced back to a straightforward genetic cause. As a
consequence, he argues that the HGP will have to be redone
for a variety of metabolic contexts in order to found a sound
technology of genetic engineering (O’Nuallain and Strohman,
2007).
In essence, O’Nuallain says, the analogy works as follows:
first of all, at the orthographic or phonological level, depend-
ing on whether the language is written or spoken, we can map
from phonetic elements to nucleotide sequence. The claim is
then made that Nature has designed highly ambiguous codes
in both cases, and left disambiguation to the context.
Here we investigate a class of probability models based on
the asymptotic limit theorems of information theory that in-
stantiate this perspective, and explore a ‘natural’ means by
which epigenetic context ‘farms’ gene expression in an inher-
ently punctuated manner via a kind of catalysis. These mod-
els will then be used to illuminate ways in which normal devel-
opmental modes can be driven into pathological trajectories
expressed as comorbid psychiatric and physical disorders, ex-
panding recent work by Wallace (2008b). With some further
work, it would appear possible to convert these models to
powerful routine tools for data analysis, much as probability
models based on the Central Limit Theorem can be converted
to parametric statistics.
1.2 The context
What we attempt is, of course, embedded in a very large
context. Jablonka and Lamb (1998), expanding on Jablonka
and Lamb (1995), have long argued that information can be