I think results depend on the tasks used, and not always tap into the quintessential elements of cognition per se. Tasks and scales used here are heavily bent on emotional systems/interface than hardcore cognitive function. These results, however, partly replicate others' findings in terms of bias to negative stimuli, inflated by depressive symptoms from residual mixed or depressive states.  Also, adult disorders with later onset, preserve cognition more than early onset bipolar disorder (relative, Clark et al, Sweeney et al., Rubenstein et al) . Which is why, patients with adult onset  show better prognosis than in case of pediatric onset where multi-system involvement is norm than exception. Again, I think large samples, life span studies, pure cognitive and affective and interface tasks in the same sample may provide ultimate answers. However, that said, good to keep these findings in mind as we navigate forwards.