Even the weak form of Sapir-Whorf is beginning to fall seriously out of favor with many linguists. A series of recent experiments which have been formatted to test concretely whether monolinguals have trouble dealing with ideas which their language does not formulate clearly seem to indicate that in fact, they do not.
One particular example I recall involved testing speakers of Chinese, which has no subjunctive mood, on their understanding of situations involving conditionality and potential action. If even the weak Sapir-Whorf hypothesis holds, it would stand to reason that monolingual Chinese-speakers would have difficulty with situations which would require the subjunctive to form clearly. The tests showed that the Chinese-speakers scored just as well, and took just as much time as speakers of English.
This has led to the formualtion and advancement of a fairly radical hypothesis in the exact opposite direction of Sapir-Whorf: that humans think in some kind of non-verbal language (to make this not a total contradiction in terms, perhaps symbolic system applies more than language), which some of its proponents have called "Mentalese". All actual thought processes, under this theory, are formulated in Mentalese, and then internally translated into the language of the internal monologue, so that they can be vocalized.
None of this, of course, is set in stone. IMHO, all of this groping about with far-fetched, contradictory, and not easilly testable theories shows just how far cognitive linguistics has to go before it reaches anything close to a definitive, or even particularly useful, state.