Models of decision making neglect the impact of emotions on information processing. Here we present a model of decision strategy selection that can account for the differences in the use of two strategies, the take-the-best heuristic (TTB) and the Weighted Additive rule (WADD) in a probabilistic inference task. How can such a model incorporate emotions? Our model assumes attentional weighting of cues, controlled by the activity of locus coeruleus – a brainstem nucleus associated with physiological arousal. Using hierarchical Bayesian modeling, the model was evaluated on data from a study where participants performed a probabilistic inference task and emotional stress was manipulated with highly aversive slides. For each participant we estimated the parameter that controls cue weighting. This parameter correlated positively with the proportion of choices consistent with TTB and negatively with the proportion of choices consistent with WADD. Moreover, this parameter was sensitive to the emotional stress manipulation.