I know a neuroscientist who explains bipolar disorder using a "sticky switch" theory of hemispheric oscillation. It's related to Ramachandran's theory of hemispheric function. Pettigrew adds to this the idea that there is a natural alternation of hemispheric dominance, on a timescale of seconds, and that in bipolar disorder, the switch rate is much slower, so that a person will spend hours or days with one hemisphere dominant. These long periods are mania and depression. He had some intriguing circumstantial evidence to do with binocular rivalry - the alternation between images was slower in people with bipolar disorder, as if this was driven by the same process.
I was looking at a hypothesis that bipolar disorder is probably due to problems with neocortical sodium-potassium pump activity cyclically decreasing and thus allowing increased resistance, which increases neuronal excitability by the square of the resistance. According to what I've read on Wikipedia it seems that agonizing Gi proteins would inhibit cAMP production and therefore downregulate sodium pump activity (this was the most tenuous part, and the reference was unintelligible) and increase neuronal excitability. But 5-HT1A (a type of Gi protein) agonists have been shown to be useful for improving symptoms of schizophrenia, which is typically thought of as resulting from increased neuronal excitability. Sign error? What's up? I don't have any model of the underlying mechanisms, and only vaguely know what words like 'downregulate' mean.
ETA: But apparently http://en.wikipedia.org/wiki/Risperidone , a 5-HT2A agonist, also supposedly mitigates symptoms of schizophrenia, which is weird because it increases neuronal excitation. It seems that the flawed assumption is that schizophrenia has something to do with 5-HT-caused neuronal excitability.
Thus it seems like maybe you can reduce bipolar tendencies by taking 5-HT1A antagonists, and schizophrenia is its own separate problem that is perhaps solved by taking a D2 antagonist. 5-HT2A agonists increase neuronal excitation which may interact somehow with the the 5-HT1A antagonists. I'm still confused.
There's this other problem where if the main reason D2 and 5-HT1A work is by decreasing neuronal excitability that might just be because sedate people are more similar to each other than manic people, but the kind of neuronal excitability that makes people manic and the kind that makes people schizoid are qualitatively different, and hence the difference in receptors being important. It could be that 5-HT2A agonists and 5-HT1A antagonists increase and decrease neuronal excitability respectively, but in different ways such that they're not countervailing.