Mental models help people navigate complex environments. This paper studies how people deal with model uncertainty. In an experiment, participants estimate a company's value, facing uncertainty about which one of two models correctly deter- mines its true value. Using a between-subjects design, we vary the degree of model complexity. Results show that in high-complexity conditions people fully neglect model uncertainty in their actions. However, their beliefs continue to reflect model uncertainty. This disconnect between beliefs and actions suggests that complexity leads to biased decision-making, while beliefs remain more nuanced. Furthermore, we show that complexity, via full uncertainty neglect, leads to higher confidence in the optimality of own actions.