Studies on the Uses and Usefulness of Diagrams
Shai Berger

Abstract:
Nr: MoL-2000-06
Title: Studies on the Uses and Usefulness of Diagrams}
Author: Shai Berger}

Cognitive Science as we know it today was built around linguistic
behavior. It was the research on language which brought down
Behaviorism in Cognitive Psychology and connected Psychology with
Linguistics and Computer Science. But as the field evolved, increasing
attention was paid to other forms of communication.

In the last two decades, we have seen a growing interest in graphics
as subject of research of different paradigms.  Graphics have been
classified by designers, their cognitive effects were studied by
psychologists, linguistic analysis has been applied to them, and
logicians and computer scientists have been studying their structural
and computational properties, all in an effort to further our
understanding of their usefulness in presenting information to humans.

There are two main motivations behind this research. The more
practical one is the idea that the principles underlying the
usefulness of graphics in communication, when well understood, can be
used to enhance the quality of presentations. The more theoretical
drive stems from the intuition that information presented graphically
is often very easy to consume --- much more so than the same
information presented verbally. Thus, understanding the principles of
graphical communication may give us important hints to the workings of
our cognitive system.

In a related, simultaneous development, the availability of graphical
tools, following the recent advances in computer technology, gave rise to many
uses of graphics as research and presentation aids, in Cognitive
Science and science in general. This may bring forth a positive
feedback loop: More use of graphics means more substance for research,
whose insights may produce better, more focused presentation
techniques, which can then be used for yet better research, and so on.

My primary motivation for this research is the theoretical one --- I
want to learn more about how people think. I believe that graphic
representations of data help people solve problems, and I want to
study how. This interest leads me towards the body of research in
diagrammatic reasoning. Within this subfield, I focussed on two,
seemingly remote, areas. 

As a computer scientist by training, I look into the computational
complexity of a typical problem encountered with diagrams in
Chapter~\ref{ch:complexity}. The problem, which we all solve on an
almost daily basis, is that of finding, extending, and verifying the
correctness of diagram legends, which are mappings of diagram features
to their meanings --- real-world entities and properties. I first
formalize this as a general computation problem. The general problem
is NP-hard, but I attempt to show that for some diagrams it is
tractable. The feasibility of solving this problem is an important
factor in the ease of use of any representation, so showing it to stem
from properties of the graphic medium goes some way towards explaining
the usefulness of these diagrams.

As a curious person with an interest in Cognitive Science, I turn my
gaze on a problem of psychological research in Chapter~\ref{ch:egg}.
The studies I look into are concerned with the effects of using
diagrams in education, for teaching logic. I expose what I believe to
be a methodological fault which made the researchers jump to some
not-entirely-warranted conclusions. In an attempt to help remedy the
problem, I suggest a new diagrammatic teaching system, and describe
its principles and prototype.

To place this research in its context, I review some previous related
work in the next chapter. I also use this review as an excuse to tackle
some philosophical issues.