"ex falso quodlibet, twin earth, philosophical method" When we reason from premisses whose truth values are unknown, we often look to their consequences in order to evaluate the truth of the premisses. (This can be done formally, as in an attempt at proof by reductio, or otherwise.) However, since false propositions imply anything (by ex falso) a mixture of true and false propositions may be inferred accidentally. This mixture has to then be culled to see if the premisses are reasonable and so on in a dialectic circle. But what happens when the premisses are *known* to be false? Take one example of a common premiss for various arguments about mental externalism, etc. in the literature. I refer to the common "twin earth" type scenarios. Assume for the sake of definiteness (if this is disputed, one can move the problem, I think, but not overcome it) that twin earth is the hypothetical locale where the oceans are not H2O (and instead "XYZ") and everything else the same. Unfortunately, twin earth is nomologically impossible - hence the situation is cannot occur, and any discussions of what things are like on this hypothetical place include falsehoods. To see that twin earth is nomologically impossible, recall some basic cell biology. On our earth, many single celled organisms feed by forming vacuoles and producing enzymes and other substances into their vacuoles when food is placed in it by various cellular processes. These cellular processes bring in the food within an aqueous medium, suitably stripped of ions and other chemical species by the cell membrane or wall, etc. Thus it is not suprising that cellular fluid consists of mostly water - dihydrogen monoxide and not "XYZ". Putnam's making up of a nonsensical chemical formula seems to obscure the fact that dihydrogen monoxide has other "lawful roles" within earth's environment, living and otherwise. Moving the XYZ to some definitive chemical species shows the radical different life that would be found on twin earth if (as we are doing) we assume the same laws apply. (I do not know what to say about anything if we assume the laws of nature are different on twin earth - such situations I will ignore because of the lack of constraints I see on our theorizing.) For example, the life on twin earth can't eat the same way we do, as I described above. It cannot form proteins by condensation reactions, as these vitally require water as the medium to receive the "extra" water produced in the reaction. If "XYZ" is the solvent of the oceans on twin earth, it thus had better be a by product of the appropriate metabolic processes as well - and then life looks *very* different. Here we enter the realm of extreme speculative science fiction. And then we have the problem associated with the laws. Thus the partisan of twin earth examples is caught in a dilemna. If he keeps them simple, as Putnam did, they involve a knowingly false premiss. If they are complicated, one has to spend one's time explaining how these creatures really are or aren't like us, and how their psychology is somehow sufficiently like ours - i.e. then one would have to defend a form of functionalism, no doubt - but in any case get mired in other philosophical concerns far from what one was originally concerned with. And, as the example of functionalism shows, we are not too far away from philosophy of mind issues the twin earth examples was supposed to illuminate, so we are in grave danger of begging the question. Finally, even if we avoid this trap, then the twin earth material is almost otiose - why not just tell the science fiction story from the get go and frame it as such and thus defend it on its merits. Moral: twin earth examples are dubious. Objections answered: Someone might rejoin that I have spoken in actualist terms about twin earth and I should speak subjunctively ("counterfactually"). The answer to this is simple. Speak subjunctively if one wants: the problem is not with the contrary-to-fact situation per se, it is that it is knowingly reasoning from a false premiss. (Well, maybe Putnam and the others don't realize that they have been, but now they do, assuming this article goes somewhere.) I suppose someone will then say that this guts all thought experiments. See the third objection. Another objection (of the form "I told you so!") would be from Belnap and the relevance logic school (or Priest and the paraconsistent guys). Yes, I see how this is grist for your mill, Belnap/Priest. But we can talk about how radical a response we need later. Third objection answered: Someone says "But thought experiments have a long and glorious history in philosophy and science. Why pick on twin earth?" Finding that premisses in thought experiments are false in some contexts leads to revising the thought experiment and its consequences as when Bohr showed that Einstein (of all people) wasn't thinking relativistically at one point in a thought experiment. Fourth, continuing from the third: But even if Einstein put in the relativistic correction in his debate with Bohr, there is still a false premiss, namely that a situation like the one described obtained. This is correct too, and shows how deep the puzzle is. My suggestion, tentative as it is, would appeal to the difference between laws and boundary conditions. The laws in the revised Einstein-Bohr thought experiments are (ex hypothesi) the ones in the real world, whereas the situation may not be realized due to conditions. I realize that getting out by appealing to what some regard as obscure (laws) is not helpful too much, but at this point I am raising a worry, but not able to solve it. Fifth: But, someone continues, it is fruitful to imagine counterfactually varying the laws in order to check our knowledge of them. Agreed. And here I get very puzzled indeed and do seem to be pulled in Belnap's direction, if not all the way to Priest's. Then I read up on what that requires and I feel it can't be done either. But that's another story for another time. Sixth: I have supposedly run together logical possibility with nomological. I reply: For one thing, logical possibility strictu sensu tells us nothing whatsoever. It is *logically* possible that meanings are in the head, and not. Why? Because logical possibility, as known for quite a while, is provability (or satisfiability, whichever - it is really two notions, which complicates matters inessentially here) within a given formal system. The sloppy pretheoretic notion is just conceivability (or so it seems - nobody has ever told me what constrains it), which for Putnam's uses is dangerous: so one can conceive of something? So what? What is supposed to follow from that? People can conceive of contradictions! It is here that a more rigorous notion of logical possibility is needed; see earlier in this section. Further, I do not buy this idea that one should care about mere logical possibility. I know of only one world, and some approximation to some of its laws (= objective patterns of being and becoming). But note also how fundamentally different the two various notions of possibility are. Sixth, bis: It seems that the appeal to nomological possibility versus logical possibility is also a red herring. Why? Because one is *still* reasoning from a false premiss. Is one supposed to simply imagine that the premisses are true? That doesn't work, or how would one ever use reductio: one would be imagining that the premisses are true, and true premisses entail true conclusions, so ... If one is allowed to *pretend* they are true and yet draw false conclusions because they *really aren't*, the problem arises again. If the conclusion is to merely be that it is conceivable that meanings ain't in the head, or that the notion is not self-contradictory or contradictory within some system or other, ok, sure, but so what then? These are extremely weak conclusions, one which are likely concludable without all the song and dance of the thought experiment. Besides, Putnam doesn't put the conclusions in such terms anyway. Seventh: I have been told sometimes to "repair" the thought experiment by allowing for a substance identical in all respects save its composition to water. And yet everything else is the same. What? How does that work? How can that scenario be even coherently described? If XYZ is identical in all its properties to water, it must react with sodium to form sodium hydroxide! And then it must contain oxygen or there's a nuclear reaction going on. Or maybe there's something exactly like oxygen propertywise but not oxygen in XYZ. And so on. Eventually one bottoms out (or goes on an infinite regress if there are infinitely many hierarchy stages to particles, strings, whatever). These enitities are exactly like their regular-earth counterparts except *what* now? Composition at the merely chemical level only delays the question. (Shorter, if you are a fan, which I am not, of supervenience: the different properties of various particles supervene on their composition.) Oh, but then my critics say, well, maybe the laws are different. Ok, but then you've gotten as far into science fiction as before. Why think one's intuitions about the psychology, etc. are at all reliable? Moreover, why repair the thought experiment at all? I think this concedes my point! If it did not matter that one was reasoning from a false premiss, why would one one need to repair at all? Eighth: One critic asked me to consider the possibility (conceivability?) that the premisses are in fact true. Ignoring the fact that this is rather extreme (and seems to gut our entire ability to use science to understand how the world is), I don't see how this helps, unless, somehow, one thinks that *every single thought experiment premiss* is in fact true (or could be determined to be true, if one has such a theory of truth).