I was reading through the SEP article on the epistemology of modality, which prompted my initial question:

Is, "Water is H2O," currently falsifiable? Or are we now in a position where, if we microscopically analyzed a packet of substance and we didn't read out H2O, we'd conclude it wasn't water?

The twater idea usually adverts to the equivalent of a parallel universe, so the issue of our biochemistry being compatible with XYZ doesn't arise, but suppose we came across some twater in our same universe. Surprisingly, even when we drink XYZ, our bodies still process it like they would water proper.

This would just be odd, I suppose, but what if we filled a bottle with equal parts H2O and XYZ? Let's say we now have a bottle of twawa. We swirl the twawa around until the bits of water and twater are uniformly distributed throughout the bottle. We could, with effort, isolate pockets of twawa and separate them back out into their constituents, but this would be more difficult than filtering out sediment, say. At any rate, why isn't the bottle of twawa, really just a bottle of water, and hence how is twater not water? Or, doesn't the bottle of twawa falsify the hypothesis that water is H2O?

Or suppose we drank a glass of what microanalysis revealed was plain H2O, and it inexplicably, but reliably, caused one of our eyes to turn into a penguin's beak. Like, everyone who drinks glasses of this water, gets a beak. Presumably there would have to be some dark matter or dark energy involved; otherwise, we would seem to have a sample of H2O that isn't water (since water proper doesn't cause beak growth in humans).

My general objection: but this really just goes to show that the meaning, not of water, but of H2O, is at issue. Basically, if there were parcels of matter that not only macroscopically behaved like water, but could be chemically melded with water proper in the bottle case described above, those parcels of matter would have to be H2O, on account of that term's own definition, not the definition of water. Identifying those parcels as H2O depends, in part, on identifying them as macroscopic water beforehand. (But so this might go on to show that the direction of definitions, in the Kripke-Putnam case, goes against the grain of how water and H2O are actually interdefined vs. actual scientific discoveries and insights.)

  • I just don’t think it’s a hypothesis that water is H2O, “doesn't the bottle of twawa falsify the hypothesis that water is H2O?” Hypothesis implies science and in the scientific description (which water is H20 is not) you’re going to specify exactly what each word is supposed to mean in the theory. If something has the chemical structure of H20 but produces beaks, there’s no great puzzle-lots of things are H20 and do different things (ice, steam, etc)
    – J Kusin
    Commented May 2, 2022 at 21:09
  • No doubt we'd tend to conclude, "Sometimes, H2O causes beak growth in humans." Or would we say, "This sample of water causes that"? It would be hard to frame, "Water is H2O," as a hypothesis, I suppose; it would have to be something like, "The words water, wet, etc. typically successfully refer relative to samples of H2O," which could be an empirical proposal about word usage; but I don't see that this would here instantly lend itself to elaborately rigid transworld identities, say. Commented May 2, 2022 at 23:01
  • I think we’d only say “sometimes, H20 cause beak growth” in the same sense of “technically, a tomato is a fruit”. If we really wanted to be talking about tomatoes when talking of fruit, we’d update our definitions. The working theory or context (culinary, botany, physics, physics post-twawa particles) changes the language.
    – J Kusin
    Commented May 2, 2022 at 23:29
  • I do know that C2H5OH is definitely not interchangeable with water. Although it can cause your nose to grow.
    – Scott Rowe
    Commented May 4, 2022 at 10:21

1 Answer 1


This question has touched on a logic error in most modal thinking, and articulates one of the test cases that illustrates that logic flaw.

The SEP article lists three sets of laws that modal reasoning postulates, laws of nature, laws of metaphysics, and laws of logic. However, there are strong reasons to think there are no absolute laws in any of these three categories. The cited SEP article in the OP, notes that the existence of metaphysical laws is problematic, and difficult to justify, so I will address the other two more plausible sets of laws.

For "nature", the "laws" of physics at least have been shown, by Noether's Theorem, to be generated by fundamental symmetries in nature. However, almost all of the symmetries we have found in nature are only regularities, not necessities. They are spontaneously broken symmetries, that hold in most cases, but not all. And our current best physics theory is that ALL of these symmetries will spontaneously break, as they are gauge symmetries, not global symmetries. https://www.pnas.org/doi/10.1073/pnas.93.25.14256 If the symmetries break, then laws would only be regularities, and every one of the laws of physics also breaks.

Additionally, for the rest of science, science is now accepted as pluralistic. See section 5 of: https://plato.stanford.edu/entries/scientific-reduction/ But a pluralist base for reasoning, CANNOT produce a coherent set of predicted outcomes. Science is therefore a pragmatic not a logic field, and does not have the internal consistency to even support "laws of nature".

For "laws of Logic", logicians have recently reached a near consensus that logic is pluralist: https://www.cambridge.org/core/journals/think/article/abs/guide-to-logical-pluralism-for-nonlogicians/EDFDFA1C9EB65DB71848DABD6B12D877 With pluralist logic, there is not a single logic conclusion to logic questions. Hence, there cannot be "laws of logic" either.

Addressing the example in the question -- is "water == H2O" a logical necessity, the argument in favor depends on both laws of logic and laws of nature being inviolable. The law of logic assumed is that of reduction, that all physical properties can be reduced to the properties of a material substrate. The laws of nature assumed is that H2O has an explicit and defined set of properties.

However, as the thought problem notes, physical properties are NOT exclusively reductionist. One can have the same set of properties, with multiple different substrates. This is the insight of "multiple realizability", that refuted neural identity theory for consciousness, and lead to the postulation of functionalism as an alternative (in functionalism, the postulated identity of consciousness was redefined to be to the function, which could be realized many different ways, not the details of a particular substrate). Similarly, water could in principle have an XYZ base chemistry, yet still be functionally "water" based on its properties.

Further discussing questions -- define water functionally and not reductively, and yes Twatter is water. So is a 50/50 mix of the two.

As a pragmatic aspect of our world, there is a large range of impurities that we always find in our H2O, yet the impure mixture still behaves as water. So our water already is multiply realizable with impurities. But SOME impurities have significant physiologic consequences on biosystems, even if at the macro level (wetting, density, freezing point) the H2O mixture might behave otherwise like water. Such poisoned water is not "water". So a sample of H2O that causes one to grow a beak out of an eye, most likely due to a trace bioactive protein catalyst, is functionally not water.

This discussion has mostly focused on the logic aspects of modal thinking incorrectly assuming laws that do not uniformly hold. One could extend this discussion to physics, where the properties of the Standard Model of Quantum Mechanics are now recognized to be contingent, not necessary. With contingent "laws" of physics possibly changing in our world over time, or between possible worlds, one could have H2O which does not behave like water.

Modal reasoning is still a useful activity, which the above discussion explains, because these "laws" are regular enough that pragmatically one is usually just fine assuming they are inviolable. But modal reasoning is only pragmatically "true", not logically true.

  • I would say that I am a methodological pluralist wittingly, but I am not sure whether I am a metaphysical pluralist. I always have to be cautious about it because if I push the ambient thought process too far, I would end up with a pluralism about the truth of monism vs. pluralism itself, which would first logically explode, and then logically implode, leaving me with a gristly monism to chew on. I suppose I'm more at the point where the very question of monism-vs.-pluralism seems perhaps incomplete or even downright inadequate to the target issues. Commented May 3, 2022 at 20:44
  • 1
    @KristianBerry -- Before I could embrace pluralism, I had to ditch logical truth, in favor of pragmatism and pragmatic truth. Yes, pragmatic systems all are subject to logic explosions, because pragmatic truth isn't capable of supporting logical consistency. But those are the only kinds of systems that can match our world.
    – Dcleve
    Commented May 3, 2022 at 20:48
  • If those are the only kinds of systems that possibly match our world, then it is necessary that no other systems match our world, and we have sailed off into the heights of modality-land again anyway, though, haven't we? I feel like this is one of those, "If every rule has exceptions, then there is an exception to the rule that every rule has exceptions, so there is still an exceptionless rule," kinds of situations. Commented May 3, 2022 at 20:55
  • @KristianBerry As an empiricist, I can't show any kind of necessity to this conclusion. It is an empirical inference. And empiricism is always uncertain, but that is pragmatically good enough.
    – Dcleve
    Commented May 3, 2022 at 21:04
  • 1
    @KristianBerry -- Currently, we don't know how to do science other than pluralistically, logic other than pluralistically, and non-science knowledge other than pluralistically. However, theory is always underdetermined by evidence, so the reasonable inference that we have to live with pluralism, could change at some time in the future. Empiricism (IE science and pragmatism), however, operates off what works from our current knowledge, not logical possibilities.
    – Dcleve
    Commented May 3, 2022 at 22:49

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .