Consciousness, cognition, and the hierarchy of context: extending the global neuronal workspace model

(7) Executive functions work as hierarchies of goal contexts.

Our particular extension of this perspective will be to in-
troduce the idea of a hierarchical structure of ‘contexts-of-
context’. We will attempt to explicitly model the roles of cul-
ture, individual developmental and community history, and
embedding sociocultural network in creating a further - and
very powerful - hierarchy of constraints to conscious events.
To do this we must bring together two other related strains
of research on neural function, cognition, and consciousness,
taken, respectively, from physics and philosophy.

The increasingly dominant global neuronal workspace
paradigm has a closely corresponding track within the physics
literature, involving adaptation of a highly mathematical
statistical mechanics formalism to explore observed phase
transition-like behavior in the brain. These efforts range from
‘bottom up’ treatments by Ingber (1982, 1992) based on in-
teracting neural network models, to the recent ‘top down’
mean-field approach of Steyn-Ross et al. (2001, 2003) which
seeks to explain empirically observed all-or-nothing effects in
general anesthesia.

Parallel to both the neuroscience and physics lines of re-
search, but absent invocation of either dynamic systems the-
ory or statistical mechanics, is what Adams (2003) has char-
acterized as ‘the informational turn in philosophy’, that is,
the application of communication theory formalism and con-
cepts to “purposive behavior, learning, pattern recognition,
and... the naturalization of mind and meaning”. One of the
first comprehensive attempts was that of Dretske (1981, 1988,
1992, 1993, 1994), whose work Adams describes as follows:

“It is not uncommon to think that information
is a commodity generated by things with minds.
Let’s say that a naturalized account puts matters
the other way around, viz. it says that minds are
things that come into being by purely natural causal
means of exploiting the information in their environ-
ments. This is the approach of Dretske as he tried
consciously to unite the cognitive sciences around
the well-understood mathematical theory of commu-
nication...”

Dretske himself (1994) writes:

“Communication theory can be interpreted as
telling one something important about the condi-
tions that are needed for the transmission of infor-
mation as ordinarily understood, about what it takes
for the transmission of semantic information. This
has tempted people... to exploit [information the-
ory] in semantic and cognitive studies, and thus in
the philosophy of mind.

...Unless there is a statistically reliable channel of
communication between [a source and a receiver]...
no signal can carry semantic information... [thus]
the channel over which the [semantic] signal arrives
[must satisfy] the appropriate statistical constraints
of communication theory.”

Here we redirect attention from the informational content
or meaning of individual symbols, i.e. the province of seman-
tics which so concerned Dretske, back to the statistical prop-
erties of long, internally-structured paths of symbols emitted
by an information source which is ‘dual’ to a cognitive process
in a particular sense. We will then adapt and modify a va-
riety of tools from statistical physics to produce dynamically
tunable punctuated or phase transition coupling between in-
teracting cognitive modules in what we claim is a highly nat-
ural manner. As Dretske so clearly saw, this approach allows
scientific inference on the necessary conditions for cognition,
and, we will show, greatly illuminates the global neuronal
workspace model of consciousness. It does so without raising
the 18th Century ghosts of noisy, distorted mechanical clocks
inherent to dynamic systems theory, and permits extension
far beyond what is possible using statistical mechanics mod-
els of neural networks.

The method opens the way for the global neuronal
workspace model to incorporate the effects of other cog-
nitive modules, for example the immune system, and em-
bedding, highly structured, social or cultural contexts that
may, although acting at slower timescales, greatly affect in-
dividual consciousness. These contexts-of-context function in
realms beyond the brain-limited concept defined by Baars and
Franklin (2003). Such extension meets profound objections to
brain-only models, for example the accusation of the ‘mereo-
logical fallacy’ by Bennett and Hacker (2003), which we will
consider in more detail below.

Before entering the formal thicket, it is important to high-
light several points.

First, information theory is notorious for providing exis-
tence theorems whose representation, to use physics jargon,
is arduous. For example, although the Shannon Coding The-
orem implied the possibility of highly efficient coding schemes
as early as 1949, it took more than forty years for practical
‘turbo codes’ to actually be constructed. The research pro-
gram we implicitly propose here is unlikely to be any less
difficult.

Second, we are invoking information theory variants of the
fundamental limit theorems of probability. These are inde-
pendent of exact mechanisms, but constrain the behavior of
those mechanisms. For example, although not all processes
involve long sums of independent stochastic variables, those
that do, regardless of the individual variable distribution, col-
lectively follow a Normal distribution as a consequence of the
Central Limit Theorem. Similarly, the games of chance in
a Las Vegas casino are all quite different, but nonetheless
the success of strategies for playing them is strongly and sys-
tematically constrained by the Martingale Theorem, regard-
less of game details. We similarly propose that languages-
on-networks and languages-that-interact, as a consequence of
the limit theorems of information theory, will inherently be
sub ject to regularities of tunable punctuation and general-
ized Onsager relations, regardless of detailed mechanisms, as
important as the latter may be.

Just as parametric statistics are imposed, at least as a first
approximation, on sometimes questionable experimental sit-
uations, relying on the robustness of the Central Limit The-
orem to carry us through, we will pursue a similar heuristic
approach here.

Finally, we invoke an obvious homology between informa-
tion source uncertainty and thermodynamic free energy den-
sity as justification for importing renormalization and gen-
eralized Onsager relation formalism to the study of cogni-

<<   1   2   3   4   5   >>

More intriguing information