Prediction Markets are a tool for group forecasting that is recently gaining popularity in companies to support decision making. In contrast to single expert judgements, group forecasts have the advan- tage of being less biased than their solitary counterparts, which have the disadvantage of being subject to singular opinions. However, some cognitive biases, such as partition dependence, still occur systematically also in Prediction Markets. In this research-in-progress we approach this problem from a Dual-process Theory perspective and argue that the market complexity is moderating information processing and therefore the occurrence of the partition dependence bias. First evidence of a pre- liminary experiment supports our hypothesis that increased complexity, introduced by a LMSR market compared to a poll, reduces partition dependence bias.