We conducted an experiment to examine whether the solution to insight probl
ems is facilitated by unnoticed hints. Our theoretical basis is a model pro
posed by Ohlsson (1992), where insight is conceptualized as a change of the
problem representation that follows an impasse. In previous studies (Knobl
ich, 1997) we found empirical evidence for two processes of representationa
l change, constraint relaxation and chunck decomposition. The present exper
iment aimed at finding out whether the representational change caused by th
ese processes can be accelerated by shortly presented hints. Two groups of
subjects were compared. The first group was presented with an extremely sho
rt hint (17 msec, masked) during problem-solving, while the second group re
ceived no hint. In phases of problem-solving, where subjects normally encou
nter impasses, hints increased solution probability selectively for tasks r
equiring constraint relaxation to be solved.