In the framework of expected utility theory, risk attitudes are entirely captured by the curvature of the utility function. In cumulative prospect theory (CPT) risk attitudes have an additional dimension: the weighting of probabilities. With this modification, one question arises naturally: since both utility and probability weighting determine the attitude towards risk, what is the relation between them? We ran a controlled laboratory experiment to answer this question. Our findings suggest that the two dimensions capture different characteristics of individual risk attitude. Though individuals who are risk averse in one dimension are likely to be risk averse in the other, the two dimensions show no significant correlation. Moreover, a significant proportion of subjects are risk averse in one dimension but risk seeking in the other.