Introduction and Overview - Computational Explorations in Cognitive Neuroscience - page 15

processing system. This chapter emphasizes that there
is really a continuum between what we have been refer-
ring to as conscious and subconscious processing.

1.6.1 Parallelism

Figure 1.4: Example of graded nature of categorical repre-
sentations: Is the middle item a cup or a bowl? It could be
either, and lies in between these two categories.

Everyone knows the old joke about not being able to
walk and chew gum at the same time. This is a sim-
ple case of processing multiple things in parallel (doing
more than one thing at the same time). In our every-
day experience, there are lots of examples of a situation
where this kind of parallel processing is evident: hav-
ing a conversation while driving or doing anything else
(cooking, eating, watching TV, etc.); hearing your name
at a cocktail party while talking to someone else (the
aptly named “cocktail party effect”); and speaking what
you are reading (reading aloud), to name just a few.
What may come as a surprise to you is that each of
the individual processes from the above examples is it-
self the product of a large number of processes working
in parallel. At the lowest level of analysis, we know that
the human brain contains something like 10 billion neu-
rons, and that each one contributes its little bit to over-
all human cognition. Thus, biologically, cognition must
emerge from the parallel operation of all these neurons.
We refer to this as parallel distributed processing (PDP)
— the processing for any given cognitive function is
distributed in parallel across a large number of indi-
vidual processing elements. This parallelism occurs at
many different levels, from brain areas to small groups
of neurons to neurons themselves.
For example, when you look at a visual scene, one
part of your brain processes the visual information to
identify what you are seeing, while another part identi-
fies where things are. Although you are not aware that
this information is being processed separately, people
who have lesions in one of these brain areas but not the
other can only do one of these things! Thus, the ap-
parently seamless and effortless way in which we view
the world is really a product of a bunch of specialized
brain areas, operating “under the hood” in a tightly co-
ordinated fashion. As this hood is being opened using
modern neuroimaging techniques, the parallelism of the
brain is becoming even more obvious, as multiple brain
areas are inevitably activated in most cognitive tasks.

Parallel processing can make it challenging to under-
stand cognition, to figure out how all these subprocesses
coordinate with each other to end up doing something
sensible as a whole. In contrast, if cognition were just
a bunch of discrete sequential steps, the task would be
much easier: just identify the steps and their sequence!
Instead, parallelism is more like the many-body prob-
lem in physics: understanding any pairwise interaction
between two things can be simple, but once you have
a number of these things all operating at the same time
and mutually influencing each other, it becomes very
difficult to figure out what is going on.
One virtue of the approach to cognition presented in
this topic is that it is based from the start on parallel
distributed processing, providing powerful mathemati-
cal and intuitive tools for understanding how collective
interactions between a large number of processing units
(i.e., neurons) can lead to something useful (i.e., cogni-
tion).

1.6.2 Gradedness

In contrast with the discrete boolean logic and bi-
nary memory representations of standard computers,
the brain is more graded and analog in nature. We
will see in the next chapter that neurons integrate infor-
mation from a large number of different input sources,
producing essentially a continuous, real valued number
that represents something like the relative strength of
these inputs (compared to other inputs it could have re-
ceived). The neuron then communicates another graded
signal (its rate of firing, or activation ) to other neu-
rons as a function of this relative strength value. These
graded signals can convey something like the extent or
degree to which something is true.

In the example in