Previous research on text-based learning has shown the relevance of hierarc
hical structures to the acquisition of complex concepts and the formation o
f knowledge structures. Because of the inflexible nature of traditional tex
t, however, these studies have been limited to comparing participants learn
ing with either hierarchical or linear presentations. As a consequence, our
understanding of the importance of hierarchies to information processing i
s only relative to that of linear text. The purpose of this investigation i
s to move beyond that comparison, to explore more deeply the relevance of h
ierarchies to information processing. For this study, the traits that chara
cterize a hierarchy were isolated and used in varying combinations to creat
e 4 different organizations for a single body of information: hierarchical,
clustered, unstructured, and linear. The creation of these structures was
made possible by hypertext technology. Participants were each assigned to s
tudy one of these systems and were then asked to take cued-association, pro
blem-solving, and factual-knowledge posttests. Results of these tests sugge
st that participants in all conditions created hierarchical representations
as they worked and that those in the nonlinear conditions used this struct
ure to guide their exploration of the material. They also suggest that an i
mportant function of hierarchies may be to define relations between concept
s. Results are discussed in relation to current theories of learning, the c
onstruction of knowledge structures, and application to educational setting
s.