Large-Scale Brain Area Functional Organization - Computational Explorations in Cognitive Neuroscience - page 214

ripherals that service various needs (e.g., the hard disk
is a memory module, the terminal is a response module,
and the keyboard is a perceptual module).
In contrast, the neural network principles summa-
rized in this chapter are fundamentally based on dis-
tributed knowledge and processing. Thus, our basic as-
sumption is that controlled processing is an emergent
property of the huge neural network that is the brain,
and is not a reflection of some kind of dualistic distinc-
tion between central and peripheral systems. Nonethe-
less, the different components of our tripartite cognitive
architecture are probably differentially involved in con-
trolled versus automatic processing. Thus, we are led to
an explanation of this distinction that is based on emer-
gent properties of the entire network, but that also incor-
porates the unique contributions that specialized brain
areas may make.
As we will see in chapter 11, the specializations of
the frontal cortex (specifically the prefrontal areas) for
rapidly updatable, robust active maintenance enable this
area to provide a sustained, top-down biasing influence
over processing elsewhere in the system. These actively
maintained frontal cortex representations can guide be-
havior according to goals or any other types of inter-
nal constraints. Thus, the frontal cortex likely plays
a more central role in cognitive control than other ar-
eas. However, it is through global constraint satisfaction
processing across all three areas (posterior and frontal
cortex and hippocampus) that representations are acti-
vated in the frontal cortex in the first place, so con-
trolled processing remains essentially distributed and
emergent. Furthermore, we emphasize the importance
of using powerful learning mechanisms to explain how
controlled behavior can be “smart,” thereby avoiding
the need for a homunculus.

the same thing as controlled processing, implying de-
liberate, conscious, intentional processing, whereas im-
plicit and procedural imply the lack of these attributes.
However, declarative also appears to imply the use of
linguistic representations, in that the information can be
“declared.” Explicit is not as tied to language, but does
seem to imply conscious access to information. Mean-
while, procedural seems to imply action-oriented be-
haviors, whereas implicit implies the lack of conscious
access.
Thus, understanding the meaning of these terms
within our cognitive architecture requires us to deal
with the issues of conscious awareness and the role
of language. Although a comprehensive discussion of
these issues is beyond our present scope, we can sum-
marize a few relevant points.
First, regarding conscious awareness, a group of re-
lated ideas in the literature focuses on the importance of
the duration, persistence, stability, and level of influence
of representations (e.g., Kinsbourne, 1997; Mathis &
Mozer, 1995). Under this general view, one can have
conscious awareness of something to the extent that
its representation has some or all of these characteris-
tics. Within a neural network perspective, this amounts
to saying that conscious awareness requires an activa-
tion pattern that is sufficiently strong to drive activa-
tions elsewhere in the network. In other words, we are
aware of those things that are playing a prominent role
in constraining the global constraint satisfaction settling
process within the brain.
As a result, one should generally have greater con-
scious awareness of things that are active in the frontal
cortex and hippocampus, as compared to the posterior
cortex. The frontal cortex and hippocampus sit in a rela-
tively powerful position at the top of the cortical hierar-
chy, and the frontal cortex in particular is specialized for
constraining (biasing) processing in task or goal appro-
priate ways. This is consistent with the idea that these
two areas also play a particularly important role in con-
trolled processing, as discussed in chapter 11.
Thus, we think there is a coherent story here that ties
together consciousness and the roles of the controlled-
processing areas of the frontal cortex and hippocampus.
Interestingly, this story does not need to make reference
to language, allowing that non-human animals can be

7.5.2 Declarative/Procedural and Explicit/Implicit

Distinctions

Another set of terms are used to denote distinctions
similar to controlled versus automatic processing, but
not completely synonymous. One such distinction
is implicit versus explicit knowledge and processing.
A related distinction is procedural versus declarative
knowledge. Both explicit and declarative connote much