Nonlinear preferences models are theories for choice under risk which
have weakened in one way or another the Independence Axiom of Expected
Utility. in this paper we present an experiment designed to discrimin
ate between some basic ideas in this literature. We tested the Between
ness axiom, a weaker form of Independence, against an alternative whic
h predicts that indifference curves in the probabilities simplex are c
onvex along the lower edge and concave along the hypotenuse. We also c
hecked whether preferences among single-stage lotteries, as those reve
aled by our test of Betweenness, should be extended to preferences amo
ng two-stage lottery via the classical Reduction of Compound Lottery a
xiom or via the Certainty Equivalent Mechanism.