Abstract In contradiction to expected utility theory, various studies find that splitting events or attributes into subevents and subattributes can reverse a decision maker’s choices. Most notably, these effects can induce first-order stochastic dominated choices. Such violations of first-order stochastic dominance are framing effects, which expected utility theory, cumulative prospect theory and salience theory of choice under risk cannot account for. However, we propose a version of salience theory which unravels the underlying mechanism triggering such effects and which can explain the impact of event- and attribute-splitting on choices. Hereby, we provide further rationale for the broad validity of the salience mechanism and its strong descriptive power concerning human decision making.