Visualizing information plays an important role in the process of teac
hing and learning. Nevertheless, there is not much information availab
le yet on how a subject matter should be visualized and which kind of
visualization is best suited for specific purposes. Traditional theori
es of text and picture comprehension do not provide an adequate basis
for answering this question. An alternative theoretical approach is de
scribed which considers comprehension of visualizations as a process o
f structure mapping between a visuo-spatial configuration and a mental
model. The empirical study carried out revealed differences in the us
e of text and picture information and in the performance in answering
different kinds of text items. These differences suggest that the stru
cture mapping process between a visualization and a mental model entai
ls not only the semantic content, but also surface characteristics of
the respective graphic display.