Caffeine is an antagonist of A2A adenosine receptors that are coupled to D2 receptors in striatally located heteromers. Antagonism on A2A has a similar effect as D2 agonism, so the dopaminergic effect (and also a psychostimulating one) of caffeine is mediated precisely by this mechanism. But is caffeine itself strong enough to cause psychosis?
Well, it would be good to know if those adenosine receptors are present in mesocortical and mesolimbic pathways as those are more implicated in psychosis. Disrupted nigrostriatal function, to my understanding, is more a consequence of the lack of regional specificity in antipsychotics. From what I've read this receptor was investigated for adjuncts to Parkinson's treatment.
Of course they are. A2A receptors are striatal postsynaptic receptors, which are also a part of the ventral striatum (medium spiny neurons in the nucleus accumbens), which is also a projection of mesolimbic pathway. Because A2A receptors are co-localized with dopamine D2 receptors, they are of course also relevant for Parkinson's disease, not just for therapy of psychoses.
I wasn't aware of A2A in the ventral striatum but would you look at that!: Article Nucleus Accumbens mediates the pronociceptive effect of slee...
In that case I fully agree that caffeine could play a role. On a similar note there have been reports of elevated urate contributing to manias and urate is an adenosine receptor antagonist. Looks like I should have spent more time reading about A2A!