The description of mental illness states brings into light a referential paradox on the absence of grounds for normality. Furthermore, the semiology itself poses a problem throughout the intricate consensual relations between psychiatrists. New molecules with activity on the CNS are ever more specific as to molecular cognitive capabilities, reaching limits of individual genetic variability. Cultural mechanisms of neuronal adaptation also contribute significantly to representations and its correlation with feelings. Neuropeptides increase excitability in various different brain regions, with networks underlying optimal behaviour patterns. Therefore, the sole specification of target molecules yet does not lead directly to specific results, as insights from a systematic approach should conceal. Current validation methods generate insufficient data for discriminating successful treatable candidates. Instead of regarding the heuristics of empirically classified disease models, a new tendency to compromise scientia rationale with technical capabilities should be regarded. Some of the drugs that have obtained patents recently will be discussed in the framework of their rational and actual specificity. The molecular basis underlining function will be contrasted with an alternative approach, namely: how functional organization constrains molecular action. The categories comprising neurogenarative pathologies at one hand and the mood disorders at the other hand will be analysed separately as the procedures guiding drug design in each case seem to be different.