Introduction and Overview - Computational Explorations in Cognitive Neuroscience - page 17

1.6.3 Interactivity

Another way in which the brain differs from a standard
serial computer is that processing doesn't just go in only
one direction at a time. Thus, not only are lots of things
happening at the same time (parallelism), but they are
also going both forward and backward too. This is
known as interactivity ,or recurrence ,or bidirectional
connectivity . Think of the brain as having hierarchically
organized processing areas, so that visual stimuli, for
example, are first processed in a very simple, low-level
way (e.g., in terms of the little oriented lines present in
the image), and then in subsequent stages more sophis-
ticated features are represented (combinations of lines,
parts, objects, configurations of objects, etc.). This is at
least approximately correct. In such a system, interac-
tivity amounts to simultaneous bottom-up and top-down
processing, where information flows from the simple
to the more complex, and also from the more complex
down to the simple. When combined with parallelism
and gradedness, interactivity leads to a satisfying solu-
tion to a number of otherwise perplexing phenomena.
For example, it was well documented by the 1970s
that people are faster and more accurate at identifying
letters in the context of words than in the context of ran-
dom letters (the word superiority effect ). This finding
was perplexing from the unidirectional serial computer
perspective: Letters must be identified before words can
be read, so how could the context of a word help in the
identification of a letter? However, the finding seems
natural within an interactive processing perspective: In-
formation from the higher word level can come back
down and affect processing at the lower letter level.
Gradedness is critical here too, because it allows weak,
first-guess estimates at the letter level to go up and ac-
tivate a first-guess at the word level, which then comes
back down and resonates with the first-guess letter es-
timates to home in on the overall representation of the
word and its letters. This explanation of the word supe-
riority effect was proposed by McClelland and Rumel-
hart (1981). Thus, interactivity is important for the
bootstrapping and multiple constraint satisfaction pro-
cesses described earlier, because it allows constraints
from all levels of processing to be used to bootstrap and
converge on a good overall solution.

Figure 1.6: Ambiguous letters can be disambiguated in the
context of words (Selfridge, 1955), demonstrating interactiv-
ity between word-level processing and letter-level processing.

There are numerous other examples of interactivity
in the psychological literature, many of which involve
stimuli that are ambiguous in isolation, but not in con-
text. A classic example is shown in figure 1.6, where the
words constrain an ambiguous stimulus to look more
like an H in one case and an A in the other.

1.6.4 Competition

The saying, “A little healthy competition can be a good
thing,” is as true for the brain as it is for other domains
like economics and evolution. In the brain, competi-
tion between neurons leads to the selection of certain
representations to become more strongly active, while
others are weakened or suppressed (e.g., in the context
of bootstrapping as described above). In analogy with
the evolutionary process, the “survival of the fittest”
idea is an important force in shaping both learning and
processing to encourage neurons to be better adapted
to particular situations, tasks, environments, and so on.
Although some have argued that this kind of competi-
tion provides a sufficient basis for learning in the brain
(Edelman, 1987), we find that it is just one of a number
of important mechanisms. Biologically, there are ex-
tensive circuits of inhibitory interneurons that provide
the mechanism for competition in the areas of the brain
most central to cognition.
Cognitively, competition is evident in the phe-
nomenon of attention , which has been most closely as-
sociated with perceptual processing, but is clearly ev-
ident in all aspects of cognition. The phenomenon of
covert spatial attention, as demonstrated by the Pos-
ner task (Posner, 1980) is a good example. Here,
one's attention is drawn to a particular region of visual
space by a cue (e.g., a little blinking bar on a computer
screen), and then another stimulus (the target )ispre-
sented shortly thereafter. The target appears either near