Spatial configuration problems can be considered as a special kind of infer
ence tasks, and can therefore be investigated within the framework of the w
ell-established mental model theory of human reasoning. Since it is a well-
known fact that content and context affects human inference, we are interes
ted to know to what extent abstract properties of linear shape curves confo
rm to previous findings of interval-based reasoning. This investigation is
done on a formally grounded basis. The main issue of this paper concerns th
e question whether the shape of linear curves in general and salient points
on the curves in particular have an influence on solving interval-based co
nfiguration problems. It has been shown in previous experiments that there
are preferred mental models if the linear structure consists of a straight
line segment. The reported experiment demonstrates under which conditions a
rbitrary shaped curves reveal similar and different effects. To distinguish
different types of points on a curve a classification of points based on o
rdering geometry is introduced. It turns out that only those shape features
are employed in solving configuration-based problems that can be character
ized on the basis of ordering geometry. Curves supplied with salient points
also lead to strongly preferred configurations corroborating the notion of
preferred mental models. Differences to the obtained types of preferred so
lutions in comparison to former investigations are discussed and possible e
xplanations are given.