This paper presents a new prototype of an intelligent learning environ
ment and the results of an experiment in learning by problem solving.
Forty-six French secondary school students solved problems of polynomi
al factorization by interacting with the learning environment during f
our forty-five-minute sessions. The intelligent learning environment i
s a new version of the APLUSIX system, called APLUSIX/MO-V2, that has
been implemented on a Macintosh computer. This system checks the valid
ity of the actions a student wants to perform, tells the student why a
query is invalid, does the calculations for the student, and gives ad
vice if asked for help. The objective of the learning experiment was t
o enable students to acquire (i) knowledge in matching formal rules of
transformation with a given expression; and (ii) heuristics of choice
between possible transformations. The results confirmed the existence
of several different levels of matching knowledge as well as several
different student learning paths in the acquisition of matching knowle
dge. Heuristics appeared to be fairly easily learned with the APLUSIX
learning environment.