Quantum mechanics does not permit parallel states to "communicate" in any way. The theory is purely linear (i.e. acts somewhat independently on all states), except for the wavefunction collapse phenomena. Any non-linenarity would, for instance, immediately imply that we can easily solve NP-complete problems using quantum computing:
So QM is probably a poor model for explaining human indecisiveness and irrationality. Much better is that it is simply the thing that evolution has come up with that can a) fit in our brains and b) yields proper results.
As far as I can see the question being addressed is which models best explain experimental results (at equivalent model complexity). If quantum(-like) probabilities pay off let's use them.
Up to now the models have mostly been a heap of heuristics but those produce models with lots of degrees of freedom that don't generalize well.
FWIW, the quantum nature of reality is harnessed by biological processes (see photosynthesis, for example[0]), and our attempts at modeling brains using classical computers have been remarkably inefficient, so far (compare the energy required by the server farm that simulates a mouse brain vs your metabolism).
I wouldn't be surprised if it turned out that cognition relied on quantum phenomena as well.
http://arxiv.org/abs/quant-ph/9801041
So QM is probably a poor model for explaining human indecisiveness and irrationality. Much better is that it is simply the thing that evolution has come up with that can a) fit in our brains and b) yields proper results.