Various hypotheses were proposed within recent years for the interpretation of the Mid Pleistocene Transition (MPT), which occurred during past 2 000 000 years (2 Myr). We here add to already existing theories on the MPT some data and model-based aspects focusing on the dynamics of the carbon cycle. We find that the average glacial/interglacial (G/IG) amplitudes in benthic δ13C derived from sediment cores in the deep Pacific ocean increased across the MPT by ~40%, while similar amplitudes in the global benthic δ18C stack LR04 increased by a factor of two over the same time interval. The global carbon cycle box model BICYCLE is used for the interpretation of these observed changes in the carbon cycle. Our simulation approach is based on regression analyses of various paleo-climatic proxies with the LR04 benthic δ18C stack over the last 740 kyr, which are then used to extrapolate changing climatic boundary conditions over the whole 2 Myr time window. The observed dynamics in benthic δ13C cannot be explained if similar relations between LR04 and the individual climate variables are assumed prior and after the MPT. According to our analysis a model-based reconstruction of G/IG amplitudes in deep Pacific δ13C before the MPT is possible if we assume a different response to the applied forcings in the Southern Ocean prior and after the MPT. This behaviour is what we call the "Southern Ocean Decoupling Hypothesis". This decoupling might potentially be caused by a different cryosphere/ocean interaction and thus changes in the deep and bottom water formation rates in the Southern Ocean before the MPT, however an understanding from first principles remains elusive. Our hypothesis is also proposing dynamics in atmospheric pCO2 over the past 2 Myr. Simulated pCO2 is varying between 180 and 260 μatm before the MPT. The consequence of our Southern Ocean Decoupling Hypothesis is that the slope in the relationship between Southern Ocean SST and atmospheric pCO2 is different before and after the MPT, something for which first indications already exist in the 800 kyr CO2 record from the EPICA Dome C ice core. We finally discuss how our findings are related to other hypotheses on the MPT.